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

From: Ben Goertzel (
Date: Wed May 03 2006 - 13:40:26 MDT


Eliezer is correct about this particular technical point of empirical
psychology: no matter how you slice it, humans do habitually make
stupid errors involving uncertainty quantification, even in very
simple contexts like the questions he cited. The kinds of objections
you have raised have been addressed pretty extensively in the
psychology literature; Eli was just citing some simple example
experiments, but there are mannnnny others addressing nearly any
possible objection.... Modern cognitive psychology may suck at
mind-theory but it's good at designing subtle experiments...

I think your point may have been that human judgment has been
evolutionarily tuned to solve different sorts of problems than these,
which is also true. Human judgment can solve problems that no current
probability-theory-based inference engine can now solve (though this
may change any year now -- just you wait!!). However, this
observation doesn't show that humans solve these problems with a
particularly high accuracy -- probably we make errors involving
uncertainty quantifications when solving the more complex and tricky
and specialized problems we have evolved to solve, as well.

My own view is that, among other strengths, the human reasoning
faculty contains very creative and clever heuristics for hypothesis
formation. We just happen to be pretty badly inaccurate at evaluating
the likelihood of the hypotheses we generate based on the available
evidence. On the other hand, existing probability-theory-based
inference engines have exactly the opposite strengths and weaknesses.

-- Ben G

On 5/3/06, Eliezer S. Yudkowsky <> wrote:
> Richard Loosemore wrote:
> >
> > Human minds are designed for immensely sophisticated forms of cognitive
> > processing, and one of these is the ability to interpret questions that
> > do not contain enough information to be fully defined (pragmatics). One
> > aspect of this process is the use of collected information about the
> > kinds of questions that are asked, including the particular kinds of
> > information left out in certain situations. Thus, in common-or-garden
> > nontechnical discourse, the question:
> >
> > Which of the following is more probable:
> > 1) Linda is a bank teller and is active in the feminist movement.
> > 2) Linda is a bank teller.
> >
> > Would quite likely be interpreted as
> >
> > Which of the following is more probable:
> > 1) Linda is a bank teller and is active in the feminist movement.
> > 2) Linda is a bank teller and NOT active in the feminist movement.
> Old, old, old alternative hypothesis disconfirmed a dozen ways from
> Tuesday. I was very quickly summarizing an extensive literature with
> thousands of papers. These are not my ideas, these are the mainstream
> conclusions of an experimental science. Go forth and read the
> literature before you make up your own interpretations. I suggest
> starting with "Judgment Under Uncertainty" and moving on to "Heuristics
> and Biases".
> --
> Eliezer S. Yudkowsky
> Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:56 MDT