by Hugo Mercier
Psychologists Nicholas DiFonzo and Prashant Bordia con-
ducted a series of studies of workplace rumors, collecting nearly
three hundred from diff er ent businesses— rumors about who
was promoted, made redundant, leaving the com pany, and the
like.14 Although accuracy varied from business to business, it was
very high: generally above 80 percent and often 100 percent. For
example, these researchers noted, “rumors about who would be
laid- off at a large com pany undergoing a radical downsizing were
totally accurate 1 week in advance of formal announcements.”15
These results replicate several older studies of the grapevine in
work environments, which all had observed rumor accuracies
above 80 percent.16
One of these studies looked at a particularly in ter est ing en-
vironment: the military during World War II.17 By contrast with
the classic study by Allport and Postman, which mostly looked
at war time rumors among U.S. civilians, psychologist Theodore
Caplow focused on rumors circulating in the U.S. Army— who
was going to be deployed where and when, who would be repa-
triated, and so forth.18 These rumors were uncannily correct.
According to Caplow: “ Every major operation, change of station,
and impor tant administrative change was accurately reported by
rumor before any official announcement had been made.”19
Some of the accurate rumors reviewed here might have made
people less anxious: soldiers hearing they would soon be repa-
triated, employees discovering they would get promoted. Un-
doubtedly, others generated significant stress: hearing one
would be sent to the front or made redundant. Whether a rumor
150 ch ap t er 10
increases or reduces anxiety has little to do with its accuracy.
What is special, then, about the contexts that consistently gen-
erate accurate rumors?
Spontaneous Rumor Tracking
At heart, the answer is quite simple: rumors tend to be accurate
when their content has significant consequences for the people
among whom they circulate.
Like any other cognitive activity, open vigilance is costly, and
we only exercise it to the extent that it is deemed worthwhile.20
This means that in domains that matter to us, we carefully keep
track of who said what, and whether what they said turned out
to be correct or not. In turn, this motivates speakers to exercise
great caution when reporting rumors, so as not to jeopardize
their own credibility.21 When we find out, eventually, whether
the rumors were true or not, our ability to track who said what
helps us create networks of reliable in for mants.
This is what enabled the U.S. soldiers studied by Caplow to
be so efficient at transmitting accurate, and only accurate, ru-
mors.22 Given the content of the rumors— such as when and
where one would be deployed—it soon became clear whether
they had been true or not. Thanks to repeated feedback, the sol-
diers learned who they could trust for what type of information,
and who should be taken out of the information network.
Moreover, for issues that relate to their immediate environ-
ment, people are generally able to check the content of rumors,
either against their existing knowledge or by gathering new in-
formation. This nips false rumors in the bud, irrespective of how
anxiogenic the situation might be.
Psychologist James Diggory studied the rumors that surrounded
an outbreak of rabies in 1952 in eastern Pennsylvania.23 People in
t i t il l at in g r um o r s 151
the most affected counties would have been the most anxious.
However, they were also less likely, compared with those in more
distant counties, to believe in exaggerated rumors about the threat.
The proximity of the threat made them more anxious but also put
them in a better position to evaluate the risks accurately.
One of the most poisonous rumors circulating in the United
States during World War II accused individuals of Japa nese
ancestry of treason, in par tic u lar of having engaged in acts of
sabotage by assisting the attack on Pearl Harbor. While these
rumors ran wild in the mainland, in Hawaii, where the sus-
pected individuals lived, they were roundly rejected “for the
people could see for themselves and could talk to the vari ous
defenders of the islands.”24
Sometimes, new prob lems arise that are practically relevant,
but about which we don’t know much, and reliable networks of
in for mants haven’t had time to crystallize. This is likely what hap-
pened during the strike at the University of Michigan. In this
novel situation, few employees had reliable information about
impor tant matters— whether classes would be canceled, whether
penalties would be imposed for striking, and so forth. The lack
of reliable prior knowledge or established networks created a rich
breeding ground for false rumors. However, because the issues
were practically impor tant for the employees, they made use of
the crisis call center created by the researchers. As a result, “in
most cases, false rumors were quelled before they could be
widely disseminated.”25
How Do We Believe in False Rumors?
Clearly, our mechanisms of open vigilance can do a very good
job when we’re assessing the majority of rumors, especially those
that affect us most. Why, then, do they seem to fail so abysmally
152 ch ap t er 10
in other cases? I argue that the diffusion of false rumors isn’t as
much of an indictment of our open vigilance mechanisms as it
seems—in fact, quite the contrary.
What is shocking when it comes to false rumors is that people
accept them on the basis of such flimsy evidence. But how do
people really believe in these rumors? Believing something— a
rumor or anything else—is not an all- or- nothing matter. Believ-
ing depends on what you do with a given piece of information.
A belief can remain essentially inert, insulated from cognitive or
behavioral consequences, if we don’t work out what inferences
or actions follow from it. Dan Sperber has called such beliefs re-
flective, by contrast with intuitive beliefs, from which we freely
draw inferences, and which we spontaneously use in grounding
our actions.26 For example, you intuitively believe there’s a book
(or other device) in front of you when you’re reading these lines.
You can grasp the book, you know you can use it to cover your
face from the sun, that you can lend it to a friend, and so on. By
contrast, take the belief that most stars you can see at night are
larger than the sun. You should be genuinely persuaded it is true,
and yet there isn’t much you can do with it.
For reflective beliefs— beliefs that tend to have fewer personal
consequences—we shouldn’t expect open vigilance mechanisms
to make as much of an effort: Why bother, if the belief doesn’t
make much of a difference? I argue that most false rumors are
held only reflective
ly, for they would have much more serious
consequences if they were held intuitively.
In some cases, it is difficult to imagine what significant be hav-
iors could follow from a rumor. Chinese citizens are hardly
going to challenge the way insurance settlements are handled in
the United States. A Pakistani shop keeper might say the Israelis
orchestrated 9/11, but what is he going to do about it?
t i t il l at in g r um o r s 153
Even when people could do something on the basis of a (false)
rumor, they most often don’t. American truthers— who believe
9/11 was an inside job— don’t act as if they intuitively believed
in the conspiracy. As journalist Jonathan Kay noted: “One of the
great ironies of the Truth movement is that its activists typically
hold their meetings in large, unsecured locations such as college
auditoriums— even as they insist that government agents will
stop at nothing to protect their conspiracy for world domination
from discovery.”27
Or take the rumeur d’Orléans, which accused Jewish shop keep-
ers of kidnapping young women. Many of the town’s inhabit-
ants spread the rumor, although for the vast majority of them,
the rumor had little or no behavioral consequences. Some young
girls started visiting other retailers, or asked friends to accom-
pany them while shopping in the suspect stores. At the height
of the rumor, some people in the busy streets stopped and stared
at the shops. Glaring is hardly an appropriate way to react after
accusations of submitting young women to a lifetime of sexual
exploitation. These be hav iors (or lack thereof) show that most
of those who spread the rumor didn’t intuitively believe in them.
By contrast, the rumors circulating in the wake of Pearl Har-
bor against Americans of Japa nese ancestry seem to have had
significant effects, as the U.S. government de cided to detain most
of these citizens in internment camps. In real ity, there were more
impor tant drivers behind the internment camps than the nasty
rumors about treason. Many of these Japa nese Americans had
been successful farmers in California, with more productive plots
than their white neighbors. Their success led to a “resentment
from white West Coast farmers,” which “provided part of the im-
petus for mass incarceration of [Americans of] Japa nese
descent.”28
154 ch ap t er 10
The lack of action following ac cep tance of the false rumors
described here suggests that open vigilance mechanisms
barely gave these rumors passing grades. If our open vigilance
mechanisms had really deemed the rumors plausible, we
should have expected altogether more power ful reactions, the
kinds of reactions we witness when people intuitively believe
in rumors.
In Pakistan, conspiracy theories about the dreaded ISI— the
intelligence service— are very common. Yet Pakistanis don’t
or ga nize conferences on how evil and power ful the ISI is. Pre-
cisely because they intuitively believe the ISI is evil and power-
ful, they don’t say so publicly.
Imagine that a female friend runs out of a shop in tears,
crying that she has been the victim of a kidnapping attempt.
Wil you be content with glaring at the vendor and, later, tel ing
other people to avoid the shop? Aren’t you instead going to
call the police immediately?
The fact that most people don’t take false rumors or conspir-
acy theories to their logical conclusion is also driven home by
the few individuals who do. Edgar Maddison Welch was one of
them. He believed the rumors saying that the basement of the
Comet Ping Pong restaurant was used by Hil ary Clinton cronies
to engage in child sex trafficking. Given this belief, coupled with
his mistrust of the corrupt police, Welch’s storming of the res-
taurant, guns ablaze, requesting the owners to free the children,
kind of made sense. Most people who endorsed the rumor—
and, according to some pol s, mil ions did— were happy doing
nothing about it or, at worst, sending insulting messages online.29
One can hardly imagine a child sex trafficker coming to see the
error of his ways as a result of reading Nation Pride’s comment-
ing on the trafficker’s “absolutely disgusting” be hav ior and giv-
ing his restaurant only one star (Google review might want to
t i t il l at in g r um o r s 155
offer the option of giving no stars for pedophile- friendly
pizzerias).30
Why did Welch take the pizzagate rumors so seriously? I hon-
estly don’t know. What matters for my argument is that of the
mil ions of people who believed the rumor, he was the only one
to act as if he did so intuitively.
Unfettered Curiosity
Even if false rumors do not, as a rule, have any serious behavioral
consequences, many people endorse them. Isn’t that a failure of
open vigilance, even if a more modest one? To understand why
this might not be a significant failure, and why people say they
believe false rumors, we must start by asking why people are in-
terested in such rumors at all. After all, if they don’t do much
with the information, why are people so keen on hearing and
spreading rumors?
Cognition is costly— a small cost for each bit of information
pro cessing, and a substantial cost for growing the brain that en-
ables it all. As a result, our minds are particularly attuned to
useful information. We come equipped with a mechanism to
recognize human faces, but not human necks.31 We are naturally
attentive to many features of potential romantic partners, but not
of programming languages. We are more interested in informa-
tion about individual humans than individual rocks.
Ideally, we should only pay attention to, pro cess, and store in-
formation that is of practical importance, information that al-
lows us to better navigate the world. However, it is impossible
to anticipate exactly which piece of information will come in
handy— indeed, attempting to make such guesses is also a cog-
nitively costly task. Your friend Aisha bores you with trivial de-
tails about her new colleague, Salma. But if you later meet Salma
156 ch ap t er 10
and get a crush on her, this information might come in handy.
Pro cessing and memorizing information is costly, but ignoring
information can be costlier, so it makes sense to err on the side
of caution, especially if the information fits a template of infor-
mation that is particularly costly to ignore.
Take face recognition. Our ability to recognize faces evolved
because it helped us interact with other people. The focus, or
proper domain—to use Dan Sperber’s terminology—of face-
recognition is composed of the faces of actual humans with whom
we interact.32 But a great many objects are picked up by our face
recognition mechanism, even though they aren’t in this mecha-
nism’s proper domain: nonhuman animal faces, a mountain on
&n
bsp; Mars, electric sockets, and so forth (figure 3; Google “pareido-
lia” for many more examples).33 This constitutes the actual do-
main of the face-recognition mechanism: all the things it can take
as input.
Why is the actual domain of our face-recognition mechanism
so much broader than its proper domain? Because of a cost asym-
metry. If you see a face in an electric socket, your friend might
find that funny; if you mistake your friend’s face for an electric
socket (or anything else, really), she will be significantly less
amused.
Mismatches between the proper and the actual domain of cog-
nitive mechanisms create vast domains of relevance: informa-
tion we find relevant irrespective of whether it has any practical
consequence. This is the source of our boundless curiosity.
Like repre sen ta tions of faces, most cultural products are suc-
cessful because we find them relevant. Celebrity gossip is an
example. If information about other people tends to be valuable,
information about popu lar, beautiful, strong, smart, dominant
individuals is even more valuable. During our evolution, we
would barely have heard of such individuals without actually
t i t il l at in g r um o r s 157
Figure 3. Two examples of pareidolia: seeing faces where there are none.
Source: NASA and grendelkhan.
interacting with them: most of the relevant information was
also practically relevant. Nowadays, we may never interact with
these salient individuals, yet we still find information about
them alluring. If you aren’t into the latest gossip about Prince
Harry and Meghan Markle, you might be interested in biogra-
phies of Lincoln or Einstein, even though you’re even less likely
to meet them than are people who read Star to meet Harry and
Meghan. Because of our interest in information about salient
individuals, we reward individuals who provide such informa-
tion, by thinking a friend who knows the latest celebrity stories
more entertaining or by buying an author’s books.
Many successful false rumors are about threats. It might seem
curious that we like thinking about threats, but it makes sense.
We may not like threats, but if there are threats, we want to know
about them. Even more than faces, information about threats
pre sents a clear cost asymmetry: ignoring information about
potential threats can be vastly costlier than paying too much
