John,
Create 10 random variables via SPSS or your favoriate statistical
package.
The distributions don't matter (for simiplicity's sake, they can all be
random
normal variate but for generality sake use a different probability
distribution
for each variable). The correlation matrix of these 10 variables will
have
a rank = 10 (i.e., cannot be reduced to a smaller matrix because the
rows
and columns are independent). This is how modules are supposed to work.
But why then do we get correlations, especially in cognitive tests?
Chomsky
might argue that for tests of language, the correlations are artifacts
of
measurement or from other sources because "the" language module
is independent of all other cognitive modules. And Chomsky will argue
until the cows come home that language is an independent module,
so take it up with him if you are feeling feisty. ;-)
Of course the real problem with "g" is that it is not theory of mind but
a mathematical consequence of factor analyzing correlation matrices.
Stop and consider: one theory of cognitive architecture for "g" is that
there is a single process that serves as the basis for thought. This
breaks down as soon as we make a distinction like short-term memory
versus long-term memory or declarative memory versus nondeclarative
memory or [insert you own favorite distinction]. What is "g" supposed
to be besides an mathematcal entity?
Or consider the following: let's call the performance of racing cars
"g" which represents winning races. All cars can be rank-ordered on
the basis of how many races and "g" explains performance. Cars
high in "g" win more races than cars low in "g". "g" is the general
ability of cars to win races. How useful is that as a concept?
NOTE: assuming "g" in this case does not require one to know
anything about automotive engineering, just how well cars perform.
Now change cars to people and races to tests. "g" is the general
ability of people to do well on tests. How useful is that as a concept?
-Mike Palij
New York University
[email protected]
----- Original Message -----
On Tue, 08 Apr 2014 11:42:06 -0700, John Kulig wrote:
It is possible that g may be modularized at the neural level, but for me
here
is the issue: we have measuring instruments that can measure g (at
least, items
that load heavily on the factor we label 'g'). This g score is usually
the best
single predictor of things like occupational success, school success,
etc.
Heritability is also highest on the g-items. Would measuring instruments
of
separate modules such as memory or specific forms of reasoning do a
better job
predicting - alone or in aggregate?
----- Original Message -----
On Tuesday, April 8, 2014 2:30:39 PM, Mike Palij wrote:
On Tue, 08 Apr 2014 05:55:37 -0700, Christopher Green wrote:
Maybe there is no g. Maybe there are independent memory and
reasoning functions but statistically they look like g because
almost all IQ test tasks require both.
http://blogs.discovermagazine.com/neuroskeptic/2012/12/24/how-intelligent-is-iq/#.U0Pwfui9KSM
Which reminds: did they ever resolve the modularity and g conundrum?
That is, if there really is such a thing like g, how does it account for
the
evidence of modularity of cognitive processes that appears to operate
independently of each other (i.e., uncorrelated)? See for example:
http://www.tandfonline.com/doi/abs/10.1080/02643294.2011.557231#.U0Q_WKLeRfQ
Really, does anyone seriously entertain "g" as a theoretical construct
and not a by-product of higher-order factor analysis?
John Kulig Tue, 08 Apr 2014 11:42:06 -0700
It is possible that g may be modularized at the neural level, but for me
here
is the issue: we have measuring instruments that can measure g (at
least, items
that load heavily on the factor we label 'g'). This g score is usually
the best
single predictor of things like occupational success, school success,
etc.
Heritability is also highest on the g-items. Would measuring instruments
of
separate modules such as memory or specific forms of reasoning do a
better job
predicting - alone or in aggregate?
==========================
John W. Kulig, Ph.D.
Professor of Psychology
Coordinator, Psychology Honors
Plymouth State University
Plymouth NH 03264
==========================
----- Original Message -----
From: "Mike Palij" <[email protected]>
To: "Teaching in the Psychological Sciences (TIPS)"
<[email protected]>
Cc: "Michael Palij" <[email protected]>
Sent: Tuesday, April 8, 2014 2:30:39 PM
Subject: [tips] How Intelligent is IQ? - Neuroskeptic |
DiscoverMagazine.com
On Tue, 08 Apr 2014 05:55:37 -0700, Christopher Green wrote:
Maybe there is no g. Maybe there are independent memory and
reasoning functions but statistically they look like g because
almost all IQ test tasks require both.
http://blogs.discovermagazine.com/neuroskeptic/2012/12/24/how-intelligent-is-iq/#.U0Pwfui9KSM
Which reminds: did they ever resolve the modularity and g conundrum?
That is, if there really is such a thing like g, how does it account for
the
evidence of modularity of cognitive processes that appears to operate
independently of each other (i.e., uncorrelated)? See for example:
http://www.tandfonline.com/doi/abs/10.1080/02643294.2011.557231#.U0Q_WKLeRfQ
Really, does anyone seriously entertain "g" as a theoretical construct
and not a by-product of higher-order factor analysis?
---
You are currently subscribed to tips as: [email protected].
To unsubscribe click here:
http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=35960
or send a blank email to
leave-35960-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
