On the Reliability of Human Witnesses
By David Talbott
Amy Acheson wrote: "But will we cultivate the
necessary discipline to find the EVENT which first gave meaning to
Finding the event behind archaic symbols is indeed the challenge.
Could any lines of reasoning be dependable, when the "evidence"
includes such enigmatic sources as myth, magical rites, and cultic
symbols? Many specialists in the hard sciences will find
abhorrent our claim that myth points to unusual natural events.
And particularly objectionable to them will be our insistence
that, under certain circumstances, human memories can give us
considerable detail about events unknown to science (Of course
the limitations of scientific knowledge come into the equation
When reports by more than one person imply a shared experience,
issues of logic and probability arise. We deal with such issues
all the time in judicial proceedings–and in fact we do not
hesitate to send someone to the electric chair based on the
memories of three people. But the principles for assessing
testimony are generally ignored when it comes to the patterns of
Rules of evidence need to be clarified, and perhaps we can work
upward from a couple of simple examples. The first question is
whether the occurrence of contradictory versions of an underlying
story excludes the possibility of a reliable reconstruction. On
this issue, common opinion is almost never correct. There are
rules for finding reliable testimony in a sea of contradictions.
Imagine an experiment involving a dozen groups with a dozen
members in each group and no communication permitted between the
groups. From each group, one individual is allowed to witness a
newly-written play, then asked to convey the story verbally to
another individual in his group, recalling as much detail as
possible. The second individual then reports to the third, and so
on until the story reaches the last person in each group, who will
then report the story to you.
From this exercise you would likely receive many different ways of
telling the story, with many contradictions between versions. But
to come as close to the original as possible you would give
greatest weight to those story elements retained in several
accounts. And despite horrendous errors in transmission within
various groups, if you follow this simple principle, your
reconstruction will be generally reliable. Even if it lacks the
full texture of the original, you can be confident in the basic structure.
To see why comparison of accounts can produce a reliable
reconstruction, you only have to recognize what a mistake in
transmission will do to a remembered event or story It will
introduce a contradiction to the way the story is told by others.
It is typically much easier to make mistakes than to make the
SAME mistakes others have made. So in the cross-referencing of
stories, the first key is to follow the points of agreement.
More significantly, there is a common paradox which even the
experts in comparative study frequently ignore. One might think
that when two groups share an improbable story element, it
becomes more likely that the two groups made the same error of
transmission. But actually the reverse is true. The more
unusual or bizarre the points of agreement, the more likely it is
that they speak for the original story. Here's why: it's much
easier to make a mistake on matters of routine background, than
on unexpected or startling detail. How many chairs were in the
room when the protagonist died? Well, there were five, but who
was counting? Here, not just mistakes, but similar mistakes would
be predictable. Consider, however, that when the protagonist
died, a dove leapt from his chest and flew away. The recurrence
of that particular element in just three of the accounts will
create a virtual certainty that the motif was part of the
original story, even in the unlikely event that the nine other
accounts failed to mention it. Short of cross contamination of
our storytelling groups, it is simply too dramatic and too
unusual to have been injected into the story by more than one
storyteller, either through a mistake or through deliberate deception.
Now these principles are extremely relevant to the cross-cultural
comparison of human memories. But there is still much more to
consider here. It is often noted that human witnesses are
notoriously unreliable. In judicial proceedings this
unreliability is properly noted–and demonstrated–all the time.
But commonly overlooked is a further consideration. In certain
circumstances the accounts of UNRELIABLE witness can produce
ABSOLUTELY RELIABLE conclusions.
To make this point I have concocted an episode called "The
Unfortunate Peter Smith"–
On Tuesday morning, a man
robbed the bank down the street,
escaping with about $12,000. When the police arrived they faced a
dilemma. The man was seen rushing from the bank toward a blue
Honda, jumping in, and speeding off. But the car was too far away
for anyone to catch the license plate.
Inside the bank, the police
found only three witnesses, and as it
turned out all were highly unreliable. One had a history of lying
relentlessly. Another was a schizophrenic, often hallucinating.
And the third was dyslexic.
Immediately on their arrival,
the police had separated the
witnesses and interviewed them. There seemed to be general
agreement that the robber was wearing a ski mask, a black leather
jacket, and blue jeans. But there were more discrepancies than
points of agreement. This was partly because the known liar
freely made up details as he answered police questions, the
schizophrenic described things seen by no one else, and the
dyslexic could not even get the name of the bank right.
Nevertheless, when the police
compared notes they immediately sent
out a bulletin, and it wasn't long before a fellow officer stopped
a blue Honda, driven by a man named Peter Smith. When the officer
looked inside the car, he did not see a ski mask, and he did not
see any money. But the moment he observed the driver, he made an
arrest. And he was certain he had nabbed the robber.
How did he know?
His confidence came from
certain details the police had noted in
their interviews with the witnesses. While much of what the
congenital liar reported was self-serving and almost certainly
invented, one thing he had said was most unusual, and was
remembered by the police interviewer. He had laughed about the
robber wearing two different running shoes. On his left foot he
was wearing a Nike, and on his right foot he was wearing an
Adidas, the man said. The second witness said nothing about the
shoes, and seems to have heard strange voices and seen things
reported by no one else. But he did mention that when the robber
started to leave the bank, several bills fell from the paper bag,
which the robber bent down to pick up. That was when the witness
noticed that the tag on his tee-shirt was on the outside; his
shirt was inside out. He could even read the label. The third
witness, the dyslexic, also had noticed the tag up close, but said
he couldn't read it. Additionally, he reported the robber wearing
two different running shoes–a Kine and a Daddies.
So the police drew a
conclusion–formulated a "prediction," if
you will–that the bank robber was driving a blue Honda, wearing
two different running shoes and a shirt inside out. And when they
found Peter Smith, they had every reason to be confident. Short
of a conspiracy to deceive them, this WAS the robber, beyond
a shadow of a doubt. THE WITNESSES DO NOT EVEN HAVE TO BE
In this example the confidence of the police relates directly to
THINGS OUT OF PLACE. A liar, a schizophrenic, and a dyslexic may
create havoc in their contradictory accounts, and yet the force of
agreement on highly unusual details is far greater than the burden
of contradictions. In fact, the convergence of testimony on the
two cited details is simply inconceivable–astronomically
improbable–unless Peter Smith was the robber. The police would
not need DNA tests, lie detector tests, fingerprints, or any other
wonders of modern science and technology to draw a reliable conclusion.
So the moral of this story is that in certain situations a simple
comparison of human testimony can achieve exceptional reliability,
even though the witnesses are not inherently trustworthy.
And how does all of this apply to the patterns of more ancient human
memory–those distinctive, archetypal complexes referring us back to the
mythical age of the gods? In this series of explorations we will
illustrate the following principles:
1) Cultures around the world,
using quite different words and
symbols, describe remarkably similar experiences;
2) These points of agreement
consistently include unique, but
well-defined forms in the sky;
3) The recurring forms have no
relationship to things seen in our
sky, or to any natural experience today;
4) Granting the presence of
these extraordinary forms will make
possible a unified explanation of myth, removing hundreds of
contradictions and anomalies left unexplained by prior theories of myth.
In seeking to reconstruct ancient memories through cross-cultural
comparison, we will discover a substructure of remarkable depth
and coherence. The power of human memory is incomparably greater
than scholars have typically assumed.