[ExI] The Anti-Flynn

Rafal Smigrodzki rafal.smigrodzki at gmail.com
Thu May 7 00:24:32 UTC 2015

On Mon, May 4, 2015 at 4:46 AM, rex <rex at nosyntax.net> wrote:

> rex <rex at nosyntax.net> [2015-05-02 10:35]:
>> Rafal Smigrodzki <rafal.smigrodzki at gmail.com> [2015-05-01 20:52]:
>>>  ### The cause of the demographic transition is still not entirely known.
>>>  The expectation of a dysgenic trend is however not dependent on the
>>> cause
>>>  of transition but rather on the verifiable negative correlation between
>>> IQ
>>>  and fitness since about 1850.
>> The high IQ fraction of the population has historically had a lower birth
>> rate
>> than the lower IQ fraction, but it does not logically follow that there
>> will
>> be a dysgenic trend. This rather astonishing result is established in the
>> paper
>> referenced at the URL.
>> http://www.nosyntax.net/cfwiki/index.php/Differential_Breeding
>> Who would have guessed that it's possible for population IQ to
>> increase while the low IQ fraction breeds at a higher rate than the
>> high IQ fraction?
> No one here finds a proof that population IQ may increase in spite
> of the dull outbreeding the sharp worthy of comment? This result
> seems patently impossible, but it's not.

### I read the article in question and it seems to make an obvious point in
the detailed part, only to use it to make statements that could be easily
overinterpreted in the discussion.

Briefly, the article uses a modeling approach to predict IQ of a population
over generations based on matrices describing the IQ outcomes of mating
between persons of different levels of IQ (the M and H functions), while
varying the number of surviving offspring (R). And yes, if you make some
assumptions about M,H, and R, you can get an increase in the fraction of
smartest individuals even if R for the smartest pairings is below
replacement and below R for less smart pairings. Once you look at the
details it becomes intuitively obvious - if you construct M and H in such a
way as to keep numbers of moderately intelligent people high, while keeping
the R for the least intelligent sufficiently low, you will get a stable
source of highly intelligent people - the average and high-average feed the
smartest fraction indefinitely even if the smartest ones do not breed well
on their own.

Preston and Campbell make the Markov assumption - they eliminate analysis
of genotypes and model exclusively phenotypes. Of course, since there was
no information on IQ genotypes in 1993, and even now there is very little,
this simplification was inevitable.This may not necessarily invalidate
their reasoning, however, it does limit the realism of their model.

The problem with their work is that it can be overinterpreted. Preston and
Campbell set out to find an explanation of the discrepancy between observed
reduced fitness of the smartest humans and the Flynn effect. They took the
Flynn effect at face value, as evidence of a true increase in IQ. They
showed that you can theoretically have a situation where despite low
fitness of high-IQ humans you could have stable or increasing average
intelligence - however, they did not show that this situation obtains in
reality. In fact, there is only a limited subset of M, H and R function
values for which average IQ remains stable or increasing, and so far I have
not found any empirical research detecting such condition in present-day

As you may recall, this thread started with an article that provides a
different explanation for the fitness/Flynn discrepancy - stating that the
Flynn effect is a superficial improvement in some intellectual parameters
(captured by IQ tests) that occurs while intelligence follows a declining
course consistent with reduced fitness of high-IQ humans.

Of course, it would be nice if Preston and Campbell's model was correct
while Woodley et al. were wrong. But, P&C only show there is theoretical
wiggle room for high IQ, while Woodley shows more direct empirical data,
with what I see as plausible assumptions. Neither article provides any
decisive data or reasoning for one of the Flynn effect interpretations -
but Woodley is for me more convincing.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20150506/b83f1783/attachment.html>

More information about the extropy-chat mailing list