by Hugo Mercier
anisms that lead to the ac cep tance, rather than the rejection, of
too many messages.
c ir c ul a r r e p o r t in g 175
Other, more mundane situations give rise to hidden depen-
dencies, situations in which many speakers appear to have come
to an agreement in de pen dently of each other, when in fact their
opinions largely stem from the same sources.
Why Do Believers Say They Believe?
We live in wonderful times: the 14th Dalai Lama is on Twitter.
His message is one of peace and toleration, including toward
other religions: “ Because of the great differences in our ways of
thinking, it is inevitable that we have diff er ent religions and
faiths.”15 This is an appealing view: we each find the religion that
best suits our needs and frame of mind. However inspirational
this statement might be, it is quite obviously false. For the vast
majority of people today, and nearly all of our ancestors, the re-
ligion they adopt is not determined by their own way of think-
ing but by where they happen to be born. People born in isolated
Amazonian tribes rarely develop a spontaneous belief in transub-
stantiation. People born in rural Pennsylvania don’t tend to
grow up believing in reincarnation.
Beliefs that are specific to dif fer ent religions— such as
transubstantiation— are pretty clearly socially transmitted. In-
deed, for most of history, just about every one adopted a version
of religion that was quite similar to that of their elders. (Note that
here I’m using a very broad definition of religious beliefs, that
encompasses beliefs in the super natural, in creation myths, and
so forth.)
Why did people accept, and why do they keep accepting,
these beliefs? People accept religious beliefs for many reasons.
One of them is simply that once every one in a community ac-
cepts some beliefs, voicing disagreement is often more trou ble
than it is worth. But there might also be a more positive reason:
176 ch ap t er 11
because every one not only believes in vari ous religious entities
but appears to have accepted these beliefs in de pen dently of each
other, hiding the dependencies between their views and making
them much more convincing in the pro cess.
The Duna, a people several thousand strong from Papua New
Guinea, believe in the kind of entities often found in traditional
human cultures: ghosts and spirits that haunt the tropical forests
the Duna inhabit. The Duna also share with many traditional
socie ties origin stories about their clans.
If the Duna’s beliefs are somewhat typical of those found in
similar socie ties across the world, their language is particularly
in ter est ing in that it contains a rich system of evidentials. When
they talk about ghosts and spirits, recount origin myths, or pro-
vide super natural explanations, they must specify the source of
their beliefs. And the sources they offer, as recorded by linguist
Lila San Roque, are revealing.16
When talking about ghosts and spirits, the Duna often use
evidentials denoting perception— saying that they have actually
seen or heard the ghosts. The Duna have an evidential form that
indicates even stronger confidence than visual perception. This
evidential is used to report things you’ve taken part in. When you
tell people, “I’ve had breakfast this morning,” you don’t leave
much room for doubt. By and large, in their everyday life, the
Duna use this evidential form as we would expect, to describe
their past actions. But they employ the same form when relating
the origin stories of their clans— which can be quite fantastical.
And so they use the same marker when saying, “I’ve had break-
fast this morning” as when they say, “Our clan’s ancestors were
birthed by ogres.”
The use of these strong evidential forms is all the more sur-
prising that Duna has an evidential form indicating “something
people say.” The Duna use this evidential when telling stories
c ir c ul a r r e p o r t in g 177
presented as fiction— the epic tales they sing for entertainment—
or rumors that invite skepticism.17 Yet, in terms of accuracy, this
evidential marker would fit their beliefs about ghosts and ogres
better than the evidentials they actually use.
Why do people tend to neglect the social sources of their re-
ligious beliefs? The complexity involved in acquiring religious
beliefs prob ably plays a role. If your only source for a belief is that
Amadou told you something, it is much easier to relay the source
accurately than when the belief stems from repeated encounters
with a group of people who share the same values and beliefs.
Another factor is that the beliefs are unlikely to be challenged.
Once a belief is broadly accepted within a community, there are
no risks in stating it as something you’re completely sure of.
Maybe more impor tant, once they have been socially acquired,
religious beliefs have some ( limited) cognitive, practical, or so-
cial consequences.
The epic tales sung by the Duna, and for which they use a
“something people say” evidential form, are entertaining, even
inspiring, but they are other wise largely inert. By contrast, a clan’s
foundation myth plays a role in justifying impor tant states of af-
fairs, such as which land a clan claims as its own. Similarly,
ghosts are used to explain misfortunes and deaths, and to justify
potential retaliation.18 Once a belief fulfil s a cognitive or social
role, it is easy to forget its social roots. A similar phenomenon
happens when we talk about microbes or Wi- Fi signals. We have
never seen either, having instead learned about them from
others. Yet they are so embedded in our actions— washing our
hands, picking a coffee shop with good Wi- Fi— and in our justi-
fications for these actions, that it is easy to forget that we owe
these beliefs to social sources (albeit, ultimately, scientific ones).
We are more likely to tell our children, “Wash your hands,
they’re full of germs” than “Scientists have discovered that
178 ch ap t er 11
there are microbes that can make people sick, and that’s why I
believe in microbes.”
While people end up accepting religious beliefs, and talking
about them as if they were things they had actually witnessed or
done, we shouldn’t forget that not all beliefs are cognitively the
same. Religious beliefs remain largely reflective rather than in-
tuitive.19 As you will recall, reflective beliefs only interact with
a limited set of inferential or action- oriented mechanisms: they
remain largely encapsulated in some part of our mind, unable to
roam free like their intuitive counter parts. Other wise, reflective
beliefs would create a lot of mayhem. For instance, the fantasti-
cal origin stories about a Duna’s clan interact with the mecha-
nisms the Duna use to understand claims about whose land
belongs to whom— but no
t with other cognitive mechanism.
A Duna might believe her ancestors were ogres, but she won’t
prepare for the eventuality that her son might be one.
What ever the reasons people have for professing religious be-
liefs as if they had acquired them on their own, how the beliefs
are presented should affect their transmission. If you grow up
surrounded by people who are competent at just about every-
thing they do, are mostly benevolent, and talk confidently of hav-
ing formed religious beliefs on their own, all cues should lead
you to accept the beliefs. Each individual testimony would have
been unconvincing (I assume you don’t believe in every god of
every religious person you have ever talked to), but the aggre-
gate makes for a very persuasive package.
What to Do?
Kaluli is another of the Papua New Guinea languages with a com-
plex evidential system. It boasts more than a dozen evidential
markers, distinguishing, for instance, between firsthand, second-
c ir c ul a r r e p o r t in g 179
hand, thirdhand, and fourthhand information. The very preci-
sion of this system made it difficult to cope with new types of
information sources brought by missionaries, such as book learn-
ing. Linguist Bambi Schieffelin chronicled the Kakuli’s efforts
to accommodate these new information sources by creating ad
hoc evidentials, one of which broadly fitted learned from a book.20
The Kakuli’s prob lem illustrates the complexity of tracing the
source of a piece of information in our modern age. When we
read an encyclopedia entry, what’s the actual source? There is the
author, but also the editor, and, more impor tant, the numerous
scholars the author relied on to write the entry, the scholars these
scholars relied on, and so forth. And the prob lem is only getting
worse— whatever the Kakuli are now doing, I hope they have a
diff er ent evidential for information found on Wikipedia and in-
formation gleaned on Facebook.21
How can we deal with this complexity? Phi los o pher Gloria
Origgi suggests that as “mature citizen of the digital age,” we
should strive to be “competent at reconstructing the reputational
path of the piece of information in question, evaluating the in-
tentions of those who circulated it, and figuring out the agendas
of those authorities that leant it credibility.” We should adopt a
reflective attitude toward sourcing, much like professionals, ask-
ing ourselves of a new piece of information “Where does it
come from? Does the source have a good reputation? Who are
the authorities who believe it? What are my reasons for defer-
ring to these authorities?”22 Doing this work would also help us
uncover hidden dependencies: as we track the provenance of dif-
fer ent opinions, we are in a better position to realize when they
all stem from the same source.
Besides doing this detective work, we should also help others
by providing accurate sources for our opinions. It is tempting to
try to get as much credit for our opinions as pos si ble. When we
180 ch ap t er 11
possess a relevant piece of information— some po liti cal news,
a scientific fact, a pertinent statistic—or even when we convey
our opinion on complex issues, we should attempt to provide
sources as accurately as pos si ble. This would often mean mini-
mizing our own role in the pro cess.23
Being more open about how we formed our beliefs not only
would help others decide for themselves whether or not they
should believe us but also would help them better understand
which sources are reliable. My guess is that if we were more ac-
curate in reporting our sources, Wikipedia, the “mainstream
media,” and other sources that are sometimes scorned would get
much more credit than they currently do, while each of us would
get a little less— and rightfully so. In turn, even if we grant less
credit to people who disclose their sources— since the ideas
aren’t their own—we should still be thankful to them, as they
improve our informational environment.
12
WITCHES’ CONFESSIONS AND
OTHER USEFUL ABSURDITIES
on november 17, 1989, the body of fifteen- year- old Angela
Correa was found in a park in Peekskil , upstate New York. She
had been raped, beaten, and strangled. Jeffrey Deskovic, a fellow
student of seventeen, reacted very emotionally to her death, at-
tracting the investigators’ attention. Brought in for questioning,
he ended up confessing to the crime.
More than a year later, Deskovic’s trial was ending. Material
evidence suggested he hadn’t done it—in par tic u lar, his DNA
did not match that of the semen found in the victim’s body. But
he had confessed, and that was enough to sway the jury, which
found him guilty. He was sentenced to fifteen years to life.1
Even though Deskovic had retracted his confession, the dis-
trict attorney (now Fox News host Judge Jeanine) refused to au-
thorize more DNA testing that might have pointed to another
culprit and exonerated Deskovic. It was only when a new DA
took office, in 2006, that more DNA tests were run, suggesting
that Steven Cunningham, who was already serving a prison sen-
tence for murder, had raped and killed Angela Correa. Cun-
ningham was convicted, Deskovic exonerated. He had spent
sixteen years in jail.
Estimating the rate of false confessions is difficult: unless a
case is overturned, it is hard to tell whether a confession was true
181
182 ch ap t er 12
or false.2 Some estimates point to a low number for minor of-
fenses: a few percent.3 Others suggest a higher proportion for
more severe offenses: more than 10 percent for people currently
imprisoned.4 What we know for sure is that false confessions are
distressingly common among people who have been later exon-
erated: between 15 and 25 percent, and even more for the most
serious crimes such as hom icide.5
Confessions, true or false, are incredibly persuasive. They
are more convincing than the most influential other form of
evidence: eyewitness testimony.6 Confessions are so defini-
tive that even when they are retracted, the accused is over-
whelmingly likely to be convicted.7 As legal scholar Charles
McCormick put it in his Handbook of the Law of Evidence, “The
introduction of a confession makes the other aspects of a trial in
court superfluous.”8
Ultimately, the relevant question here is: Why do our mecha-
nisms of open vigilance seem to be blind to the possibility of
false confession? But I must first address what seems to be an
even more puzzling question: Why do people confess to crimes
they haven’t committed?
False confessions are offered for a variety of reasons. Many
voluntary false confessions are given to cover for someone else.9
Some have weirder motives, from impressing one’s partner to
hiding an affair.10 But most false confessio
ns aren’t fully volun-
tary. Instead, they are coerced through a variety of means, rang-
ing from old-school physical abuse, to raising expectations of
leniency, or even promising small, immediate rewards, “being
allowed to sleep, eat, make a phone call, go home.”11 In some
jurisdictions— such as some U.S. states— the interrogators can
also lie to suspects, telling them that the police hold over-
whelming evidence against them. By skil fully highlighting the
short- term gains, while lessening the long- term costs, investigators
w i t c he s ’ c o nf e s si o ns 183
can make it seem like a confession is a suspect’s best option,
even if they haven’t done anything wrong. When Jeffrey Des-
kovic confessed, he had been, by his account, yel ed at for hours,
threatened with the death penalty, and told that if he confessed
the abuse would stop and he would be sent to a mental hospi-
tal.12 That was enough to break an emotionally fragile teenager
(indeed, false confessions overwhelmingly come from young
and/or mentally challenged people).13
What ever the reasons, false confessions are routinely pro-
duced, and are then generally believed. For our mechanisms of
open vigilance, a confession ticks all the boxes for a reliable mes-
sage. First, speakers are supposed to be competent, as they sim-
ply have to report on things they have done themselves. We
might understand that someone doesn’t remember doing some-
thing they have in fact done, but to falsely remember doing
something (often horrible) that they haven’t done stretches the
imagination. Second, speakers are expected to be honest. Our
sensitivity to speakers’ incentives means that self- interested
statements— such as denials of wrongdoing— are sure to be
heavi ly discounted, but also that self- incriminating statements
are easily accepted.
To make things worse, people are just as incapable of discrimi-
nating between true and false confessions, on the basis of the
accused’s demeanor, as they are between truths and lies more
generally (for reasons explained in chapter 6). The police and
laypeople perform equally poorly in this re spect, the only dif-
ference being that professional interrogators are much more
confident in their abilities, even though this confidence is
