Re: Chunking intelligence functions (was Re: [Fwd: com-mensa-rate digest#1] )

From: Eliezer S. Yudkowsky (
Date: Wed May 09 2001 - 13:34:08 MDT

Damien Broderick wrote:
> At 03:42 AM 5/9/01 -0400, Eliezer wrote:
> >Intelligence *is* compartmentalized. But not into
> >convenient, heartwarming seven-item lists.
> Wellll.... Why `heartwarming', hmmm? Why `convenient'? Could it be because
> we have evolved to chunk our abilities in this way,

Okay, I see the point of dispute. I strongly doubt that we've evolved to
chunk our abilities a certain way, or indeed to chunk them any way at all,
but that's not the point. You can obviously come up with a metric for any
variety of "intelligence" you like, from plumbing intelligence to
Broderick-novel-plot-recall-intelligence. But the true, underlying
components of intelligence - the ones that correspond to hardware - are
neither heartwarming, nor convenient.

> and to recognize them
> in action, just as we have with colors out of a continuous spectrum?

One of the reasons we know that color recognition is evolved is that all
languages follow a certain hierarchy of color distinctions and, within
that hierarchy, they all name the same colors. The eleven basic color
categories - black, white, red, yellow, green, blue, brown, purple, pink,
orange, and gray - are not all within all languages, but in those
languages where they are present, they always refer to the same color
prototypes; and in those languages that have fewer basic categories, the
basic categories are always some union of the basic terms listed. When a
language has only two basic color terms, they are "black" and "white", or
rather "cool" (black, blue, green, gray) and "warm" (white, yellow,
orange, red). When a language has three basic color terms, they are
black, white, red. When a language has four basic color terms, the fourth
is one of the following: yellow, blue, green.

The actual hierarchy is:

black, white
yellow, blue, green
purple, pink, orange, gray

A language that has a word for pink will always have words for brown,
green, and red.

("Basic Color Terms", Brent Berlin and Paul Kay, 1969. As summarized in
"Women, Fire, and Dangerous Things: What Categories Reveal about the
Mind" by George Lakoff.)

Demonstrate a similar set of language-level distinctions for intelligence
categories and your name will live forever.

> I think just for the moment I'll go along with Howard Gardner and Jerome
> Bruner and those guys who've been working in this area for the last few
> decades, and take Eliezer's guess as an interesting `sidewalk supervisor'
> opinion.

Heh. Well, that's your decision. Just bear in mind that where an
academic field has been corrupted by political influences, it's very easy
for people who have been working there for "decades" to be totally,
flagrantly, utterly wrong. I'd trust an amateur reader of "The Adapted
Mind" over any academic trained in the "soft" social sciences when it
comes to knowing, for example, "whether the brain has domain specificity
or is a general learning machine", because it is known that the standard
social sciences model is blatantly untrue, for political reasons, with
respect to that particular question.

-- -- -- -- --
Eliezer S. Yudkowsky
Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2b30 : Mon May 28 2001 - 10:00:03 MDT