Re: The Conjunction Fallacy Fallacy [WAS Re: Anti-singularity spam.]

From: Eliezer S. Yudkowsky (
Date: Tue Aug 29 2006 - 20:08:26 MDT

Jeff Medina wrote:
> Paring away the ad hominem that came later, and the admittedly unusual
> fact that Eliezer had to look up Johnson-Laird (though there's a
> difference between having to look him up because he didn't know who he
> was or what he did generally, and having to look up his large
> bibliography of work to verify whether or not J-L did any work on
> H&B), you two have just been talking past one another.

FYI: The name "Johnson-Laird" sounded familiar to me - I might have
picked it to check for that reason, I don't recall - but in general I'm
not good with names. Or faces, for that matter. I can tell they're
familiar but not what identity they should associate to.

Maybe the most embarassing case of this: I once had dinner with an old
email friend, spent around four hours with him, and it wasn't the first
time we'd ever met in person either - we'd met at least half a dozen
times over the years. And then, less than a month later if I recall
correctly, I was looking, puzzled, at someone at a local Alcor
meeting... I was *sure* I knew that face...

Sometimes I do feel very much like a stereotypical Absent-Minded Professor.

More importantly, I am not so foolish as to think that the field of
cognitive psychology is small enough that any one human being could
reasonably be familiar with all of it. I do not apologize for failing
to recognize Johnson-Laird's name. If I were a specialist in
cognitive psychology, rather than Artificial General Intelligence which
draws on a dozen other disciplines (currently learning: nonclassical
mathematical logic), it would be embarassing. If I'd made a
pinpointable stupid error (in one of my own predictions, thank you, not
my alleged interpretation of someone else on a mailing list, etc.) due
to my unfamiliarity with widely known theories originally propounded by
Johnson-Laird, that would be embarassing.

It's nice that I can associate the name "Rosch" to prototype theories of
categorization, but the useful thing is to be familiar with the
prototype theories of categorization, whether I read them in Rosch or
elsewhere, and whether I associate them to Rosch's name or not.

Johnson-Laird's ideas certainly sounded very familiar when I looked them
up, but I doubt I could have identified who "Johnson-Laird" was without
Google. Such is the truth.

Very recently I learned the Z*deh interpretation, a nonclassical logic.
  Any academic specialist in nonclassical logics would recognize the
name on sight, I'm sure, and readily fill in the missing letter. But I
wouldn't expect, say, Peter Voss to know it (though he might), and not
because I disrespect Peter Voss. There is far more human knowledge than
any one human can learn. One month ago I would have recognized the name
as vaguely familiar without being able to say which field it associated
to; and in a year I may well have forgotten the name, though not the logic.

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

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:57 MDT