Re: Is IQ usefully predictive? (not in one case)

From: Ken Clements (Ken@Innovation-On-Demand.com)
Date: Sat Aug 25 2001 - 23:06:09 MDT


My favorite IQ story happened to a friend of mine who's name is David.
Back in
the late 60's David was a student in Philadelphia looking for a summer
job in
the publishing industry. The odds of finding such a job at the time
were very
small, but David was a determined young lad, so he hit the pavement and
started
knocking on doors.

One day, David went to a company where he was told (as usual) that there
were no
openings. David asked to apply anyway, just in case. So they gave him
an
application, which included a 50 question flash IQ test to be completed
in only
3 minutes. To their complete shock, David returned the test to the HR
people in
1 and 1/2 minutes with all 50 questioned answered. They asked him
"Don't you
want to check your answers while you still have time?" and he replied
"No, they
are all done, and they are all correct." When they graded the test they
found
that they were all correct, and figured that he was some kind of boy
genus, so
they hired him for the summer thinking that, surely there was something
really
good he could do for them. And he did, and all ended very well, but
what the HR
people did not know was that David had taken that same test 6 times in
the
previous 4 days.

Another true story I found important was a case history of prefrontal
brain
damage in one of William Calvin's books. The patient was a intelligent
lawyer
with a big firm, but after his injury, he just was unable to do anything
productive. The key point is that they tested his IQ after the injury
and found
that it was just the same as before. This shows that it is possible for
IQ to
be useless if not supported by other stuff going on in the brain, and
indicates
that for a given individual (n=1) IQ may have no predictive value.
History is
full of cases of people who could solve puzzles if you sat them down and
stuck
the puzzle under their noses, but otherwise were lost to the world.

A few years ago I decided to try out some IQ tests to see how I scored.
I did
pretty well, but not as well as I had expected. So, I looked carefully
at the
design of the tests (these were general population tests) and at the
kinds of
errors I was making. I found that I had three problem areas; the first
was
anagrams. All the tests had anagrams which my brain just would not do.
I was
never interested in them in the first place, having a hard enough time
with
spelling in general, and considered the entire subject stupid because
you can
solve them easily by putting the letters in alphabetical order and
looking up
the equivalence group in an anagram dictionary. Why learn what you can
do in a
lookup? Well, to get a higher score on these stupid tests, that's why.
So I
did. As a side benefit, I can now play Scrabble with my family.

The second problem I had was with logic. That came as a big surprise to
me
because I was an A student in mathematical logic in college, and had all
these
years of designing logic circuits and writing computer code. The
problem was
that I was using very precise mathematical logic to answer questions
that were
being put to the general population. So, I would be too picky and
answer "none
of the above" when one of the answers was mostly correct but had flaws
(i.e. if
you designed circuits or wrote code that way your plane would crash, but
so
what). Therefore, I learned to ask myself "How would a logical person
who is
not technical answer this question?" and got them all after that.

The last problem was similar, but more subtle. Sometimes when given the
usual A
is to B as C is to __ question I would see the simple relationship and
then see
a deeper connection that indicated a different answer. At first I was
troubled
thinking that this was a trap for people jumping to conclusions, and I
was not
going to fall for it. Wrong, I was finding stuff that was not there.
The
answer was to just give the simple first thing out.

I mentioned all of the above because making these simple changes in the
way I
was taking the tests added 20 points to my IQ. This was an artifact of
the
general population tests; perhaps this happened to Feynman. When I
looked at
some of the "high difficulty" tests, none of the above problems
existed. This
has led me to "Ken's Theory of General IQ Tests" which goes like this:

If you submit a group of progressively more mentally capable people to
general
IQ testing you will see that the IQ goes up with capability for a bit
after 100,
but then levels off, or even starts going down. This is the point where
the
people are smarter then the test writers. Then after a bit more, the
scores
will start going up again. This is the point were the people know that
they are
smarter than the test writers, and start feeding the test "what it
wants." I
suspect that if you keep going you will get to a total drop off because
you hit
a population so smart they refuse to take the test at all, or use it for
scratch
paper while thinking about their own projects.

My view is that, except for some special cases in children where you can
use IQ
to tell who needs special help, IQ is bunk, especially in adults. Adult
brains
have many different modules that think about things in different ways.
It is a
multidimensional space that cannot be mapped on a simple Gaussian
distribution,
although many tiny functions, individually, can be. J. R., put it on
the list.

-Ken



This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:40:14 MDT