Storytelling, a follow-up -Reply

[Alan Gunn]

In message Wed, 20 Aug 1997 19:02:10 -0500,
  Jack Balkin <jbalkin at MAIL.LAW.YALE.EDU>  writes (in part):

> I'm adverting to an
> important fact about how human beings process information.  People
> tend to experience events in terms of patterns of similar causal
> sequences of events. . . .

> [S]tories are persuasive because they present us with an
> exemplar or a prototype or pattern of a causal sequence of events.
> Because people know that events generally fall into such narrative or
> causal patterns, evidence of one example of a sequence leads us to
> believe that it is an example of a larger pattern of related sequences of
> events.  This inference may be unjustified, but it is justified enough times
> so that the use of salient examples is a useful heuristic for decision.

   Many widespread beliefs that are demonstrably untrue are caused by just
this kind of inference: "dangerous" would be a more-accurate word than
"useful" in the last-quoted sentence.  Many studies of rationality show that
people often err because they look for causality behind patterns that
reflect only random variations.  For example, many (maybe most) basketball
players and fans believe that there is such a thing as "hot hands"; that is,
they believe that a player who has sunk several shots in a row is more
likely to make the next shot than if he had just missed a shot.  An analysis
of the shooting records of the only NBA team to keep complete records shows
that this is not true.  But people see it happen, forget (if they ever
really knew) that streaks are likely from time to time even for things like
flipping coins, and assume a causal relationship when none exists.  This
phenomenon accounts for such things as belief in psychics (people remember
the times they guess right and forget the others) and in things like "more
babies are born when the moon is full than at other times."
   There's a wonderful and very accessible book on this: Thomas Gilovich's
"How We Know What Isn't So."  Gilovich reports an interesting study of the
effects of various kinds of schooling on students' ability to tell the
difference between data that really tend to support something and data that
merely reflect random fluctuations.  The study looked at education in law,
chemistry, psychology, and medicine.  A couple of years of psychology did a
lot for students in this area, medicine did some good, law and chemistry
produced no improvement at all.  The ability to tell whether something
has or hasn't been shown to work is important in dealing intelligently with
social problems--Gilovich's study involved data purporting to show that a
particular program had reduced crime.  Apparently, legal education does
nothing to make students good at this.  Why lawyers think that they have
*any* advantage over anybody else in coming up with the "right answers" to
social problems is one of the great mysteries of life.  (When I teach torts,
a field which has become almost a joke because of judges' beliefs that they
know how to make the world safer, fairer, more efficient, or whatever, I
stress this: my students hate it.)  Here's just one example: economic theory
predicts that safety regulations will not increase safety.  Many studies of
things like auto-safety regulation and requirements that some medicines be
available only if prescribed by a doctor confirm this.  How many lawyers
know this, or care?
   Of course, it must be the case that *some* individual stories reflect
causality, and inferences from those stories would be accurate.  It is just
as true that flipping a coin will give the right answer to some questions,
some of the time.  The trouble with story-telling, as with coin-flipping, is
that one can never tell which story or which coin toss is right without
looking at something else.

<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
Alan Gunn
Notre Dame Law School