Thursday, 6 March 2014

Comment to Greg Cochran on the decline of intelligence since Victorian times


Greg Cochran has been the most significant (intellectually substantial) critic and opponent of the idea (deriving from myself and Michael A Woodley) that historical reaction time data have shown a significant (approx. one standard deviation or 15 modern IQ point) decline in intelligence since Victorian times. 


In his latest blog posting, Greg takes another side swipe at the idea.

Here is my comment in response.



As you presumably know, I have an extremely high regard for your work (e.g. having provided a back page blurb for 10,000 Year Explosion and invited you to write for Medical Hypotheses on the germ theory of male homosexuality).

And I am – on the whole! – grateful for your opposition to the finding of an approximately 1SD (15plus IQ points by modern measurements) decline in general intelligence in England (and similar places) as measured by simple reaction times since about 150-200 years ago – grateful because it has stimulated me to organize my thoughts on the subject.

But I continue to think you are wrong! and that the evidence you bring against this decline is inadequate – so I continue to hope to persuade you otherwise.

I have three considerations to offer.


1.       The decline in question is (roughly) from IQ 115 to IQ 100 over the space of 150 years – about one IQ point per decade (whatever that means!). But I suggest that this would not be expected to have analogous functional consequences to a decline from 100 to 85, since IQ is not an interval scale.

(In a nutshell, I think Victorian English IQ was *about* the same or a little more than recent Ashkenazi IQ – but has declined.)

This 150 year decline measure in modern IQ units corresponds to a slowing of simple reaction times from approximately 180 to 250ms for men – about 70 milliseconds.

And the minimum RT in the Victorian studies was about 150 ms – which is probably near the physiological minimum RT (and maximum real underlying IQ) constrained by the rate of nerve transmission, length of nerves, speed of synapse etc.

So average Victorian RT was about 30 ms above minimum RT, while modern RT is about 100 ms above minimum.

By contrast – modern reaction times (in Silverman’s study) for men average approximately 250ms with a standard deviation of 50ms – however there are good recent studies with an average RT of 300ms for men.

I would argue (on theoretical grounds) that as RT declines there ‘must’ come a point when it comes-up-against the neural constraints of intelligence, such as short term/ ‘working’ memory (the metal ‘workspace’, activation of which lasts a few seconds, seemingly) – and therefore there would be a non-linear effect of reducing intelligence – intelligence would cross a line and fall off a cliff.

My assumption is that a reduction in (modern normed) IQ from average 115 to 100 would *not* have such a catastrophic effect on high level intellectual (abstract, systemizing) performance as a reduction from average 100 to 85. (At a modern average IQ of 85, top level intellectual activity is *almost* entirely eliminated.)

When we are dealing with the intellectual elites, the same may be more apparent – the initial reduction in RT may retain the possibility of complex inner reasoning; while after a certain threshold the number of possible operations in the mental workspace would drop below the minimum needed for high scale intellectual operations.


2.       It may be that your example of maths does not refute the observation of reduced intelligence. It may be that modern mathematic breakthroughs are of a different character than breakthroughs of the past – and do not require such high intelligence.

I think this may be correct in the sense that I get the impression that modern maths seems to be substantially a cumulative, applied science – somewhat akin to engineering in the sense of bringing to bear already existing techniques to solve difficult problems.

So a top level modern mathematician has (I understand) spent many years of intensive effort learning a toolbox of often-recently-devised methods, and becoming adept at applying them, and learning by experience (and inspiration) where and how to apply them.

This seems more like the Kuhnian idea of Normal Science than the Revolutionary Science of the past – more like an incremental and accumulative social process, than the individualistic, radical re-writings and fresh starts of previous generations. And, relevantly, a method which does not require such great intelligence.

I also note that many other sciences, from biology to physics, have observed the near-disappearance of individual creative genius over the past 150 years – and especially obviously with people born in the past 50 or so years - which would be consistent with reducing intelligence.


3.       Michael Woodley and I have discovered further independent – but convergent – evidence consistent with about 1 SD (15 IQ point) decline in intelligence from Victorian times, again using simple reaction time data – but, as I say, using a completely different sample and methods. The paper is currently under submission.

I mention it because the unchallenged consensus post-Galton has been that simple reaction times has some causal – although not direct – relationship to intelligence; and if we have indeed established that RT has substantially slowed over recent generations, then either this would need to be acknowledged as implying a similarly substantial decline in intelligence – or else the post-Galton consensus of IQ depending on RT would need to be overturned.


Note added: An e-mail correspondent writes:

Take the following claim of Cochran's:

"In another application – if the average genetic IQ potential had decreased by a standard deviation since Victorian times, the number of individuals with the ability to develop new, difficult, and interesting results in higher mathematics would have crashed, bring [sic] such developments to a screeching halt. Of course that has not happened."

Cochran is completely correct in his reasoning, and in his prediction that higher mathematics would have crashed given a one sigma decline in g. His last sentence is however empirically false, because a crash is precisely what the data indicate happened.

Charles Murray, in his 2003 Human Accomplishment presents graphic data of the rate of eminent mathematicians and major accomplishments in mathematics (p. 313). The trends reveal a precipitous decline in the occurrences of both of these between the years 1825 and 1950. Extrapolating the decline in this period out to the year 2000 would place the rate of eminent mathematicians and their accomplishments below the rate observed in 1400, despite massive population growth in the West during this interval. The peak of mathematical accomplishment clearly occurred during the heyday of eugenic fertility in the West, between 1650 and 1800, and actually occurred earlier than the peaks experienced in other areas of science and technology, perhaps suggesting greater sensitivity to shifting population levels of g (a testable prediction incidentally).

These data completely concur with my sense that modern 'mathematics' has stagnated. There are virtually no valid proofs being offered for the long-standing mathematical problems these days. Six of the seven Millennial prize problems remain unsolved. More worrying still, no one seems to have grasped the enormity of the problem posed to the foundations of mathematics by Georg Cantor's work on transfinite numbers, and we are no closer to understanding how these fit into the foundations of mathematics today than we were in the 1900's.

The two greatest mathematicians alive today are Andrew Wiles, who solved Fermat's Last Theorem, and Grigori Perelman, who amongst other things, solved Poincare's Conjecture (the only Millennial prize problem to have been unambiguously solved thus far). Of the two of these, Perelman is the only one who would compare favorably with the great mathematicians of the past. Wiles, whilst having undoubtedly made a major discovery, is clearly second rate by historical standards, as he had to marshal enormous amounts of time and effort into solving just one problem, which was not completed until he was more than 40 years old - an achievement pattern atypical of great mathematicians who typically reach peak accomplishment at less than 35 years of age.

That leaves Perelman, who has been prodigious and productive from a  relatively early age. He is nothing if not scathing about the state of modern mathematics either, having claimed the following in a 2006 interview on why he turned down various prestigious mathematics prizes:

"Of course, there are many mathematicians who are more or less honest. But almost all of them are conformists. They are more or less honest, but they tolerate those who are not honest."

This could of course equally well apply to every area of scientific inquiry in the modern world. Data, such as that presented by Murray and others clearly reveal that what you have today are hoards of 'mathematicians' who are collectively not one iota as accomplished as the relatively less numerous, but vastly more talented individuals who dominated this field in centuries past.

Just because these over-promoted self-promoters claim something is 'interesting', 'new' or even a 'breakthrough' in their field doesn't make it so - the decline in eminence in point of fact makes it antecedently highly implausible that 'mathematicians' today are even capable of generating anything approaching a breakthrough (ultra-rare individuals such as Perelman and Wiles excepted).