by Hugo Mercier
If one side is endowed with specific adaptations, either to emit
or to receive some information, without a counterpart on the
other side, there is no genuine communication. Instead of com-
munication, there can be cues, which only require adaptations
on the receiving side. For instance, adult mammals can differen-
tiate babies from adult members of their species. But they do
not need communication to do so; they can rely on cues— most
v i g il a n c e in c o mmuni c at i o n 19
obviously, size. Babies did not evolve to be small so that they
could be recognized as babies. Small size is a cue to babyhood,
not a signal.
Now, the theory of evolution by natu ral se lection dictates that
if communication mechanisms have evolved, they must have done
so because they increased the fitness of both the entities that send
signals and the entities that receive signals. Fitness, in evolutionary
theory, is the reproductive success of an entity— which includes
not only its own reproduction but also the reproduction of copies
of itself. So individuals can increase their fitness by having more
descendants, but also by helping their kin, who are more likely to
share any new gene variant the individual possesses, have more
descendants— what biologists call inclusive fitness.
In some cases, the evolution of communication is straightfor-
ward enough. The cel s of an individual share the same fitness:
the cel s of your liver and your brain both increase their fitness
when you reproduce. Their interests are perfectly aligned with
each other’s. As a result, there is no reason for a cell to mistrust
what another cell from the same body might communicate, no
obstacle to the evolution of communication between them. In-
deed, our cel s keep listening to each other even when some of
them turn bad: cancerous cel s emit signals that tell the body to
grow more blood vessels, and the body obeys.7
Entities can also share the same fitness without being part of
a single body. The fitness of worker bees, for instance, is entirely
tied with the reproductive success of the queen. The workers can-
not reproduce on their own, and their only chance of passing
on their genes is through the queen’s offspring. As a result, worker
bees have no incentive to deceive one another, and this is why
a bee can trust what another bee signals without having to
double- check— even if the bee suggests there are flowers in the
middle of a lake.
20 ch ap t er 2
Still, a lot of communication happens between individuals
that don’t share the same fitness. In these potentially conflicting
interactions, many signals might improve the fitness of senders,
while doing nothing for the fitness of receivers, or even decreas-
ing their fitness. For instance, a vervet monkey might give out
an alarm call not because there is a predator in sight but because
it has spotted a tree laden with ripe fruit and wants to distract the
other monkeys while it gorges on the treat. We might refer to
such signals as dishonest or unreliable signals, meaning that they
are harmful to the receivers.
Unreliable signals, if they proliferate, threaten the stability of
communication. If receivers stop benefiting from communica-
tion, they evolve to stop paying attention to the signals. Not pay-
ing attention to something is easily done. If a given structure is
no longer advantageous, it dis appears—as did moles’ eyes and
dolphins’ fin gers. The same would apply to, say, the part of our
ears and brains dedicated to pro cessing auditory messages, if
these messages were, on balance, harmful to us.
Likewise, if receivers managed to take advantage of senders’
signals to the point that the senders stopped benefiting from
communication, the senders would gradually evolve to stop
emitting the signals.8 Communication between individuals that
do not share the same incentives— the same fitness—is intrin-
sically fragile. And the individuals don’t have to be archenemies
for the situation to degenerate.
Surprising Communication Failures
We tend to think of pregnancy as a symbiotic relationship be-
tween the mother and her offspring. In fact this relation is, to
some extent, conflicting from the start. To maximize their own
fitness, mothers should not dedicate all of their resources to the
v i g il a n c e in c o mmuni c at i o n 21
fetus they are currently carry ing. Instead, some resources should
be devoted to past and future children (and thus to the mother
herself). By contrast, the fetus should evolve to be biased toward
itself, compared to its siblings. As a result of this asymmetry in
the selective pressures bearing on mothers and their fetuses, the
fetuses should evolve to ask more resources of the mothers than
the mothers would optimally allocate to any one offspring.
David Haig, a particularly inventive evolutionary biologist,
suggested that this difference of selective pressures between
mothers and fetuses explains, among many other phenomena,
the strangeness of pregnant mothers’ insulin physiology.9
Through the placenta, the fetus produces and releases hormones
into the mother’s blood. One of these hormones, human placen-
tal lactogen (hPL), increases insulin re sis tance. The more resis-
tant the mother is to insulin, the longer her blood sugar remains
elevated, and the more resources the fetus can grab. In response,
the mother increases her production of insulin. At the end
mother and fetus reach a kind of equilibrium in which blood
sugar remains elevated slightly longer than usual, but much less
longer than would be the case if the mother didn’t release
increased doses of insulin. The efforts made by the fetus to ma-
nipulate its mother’s blood sugar levels are staggering: the pla-
centa secretes one to three grams of hPL per day.10 For a tiny
organism that should be busy growing, this is a significant expen-
diture of resources. By comparison, placental hormones that are
not subject to this tug- of- war can affect the mother with doses a
thousand times smaller.
If evolutionary logic makes sense of some bizarre phenom-
ena, like a mother and her fetus using hormones to battle over
resources, it also opens up new dilemmas. Take alarm calls. Until
the 1960s, their function was pretty much taken for granted: an
individual gives these calls to warn its fellow group members. The
22 ch ap t er 2
belief was that even if giving alarm calls means spending time on
the lookout rather than, say, feeding, as well as being more vul-
nerable to predation, it is worthwhile, since it increases the odds
that the group will survive. In his classic Adaptation and Natu ral
Se lection, published in 1966, biologist George Wil iams forcefully
argued against this logic. Imagine that one individual in the group
evolves to not give alarm calls, or to give them less often. This
individual is better off than every one else: it sti
ll benefits from
the others’ warnings but pays a lesser cost, or no cost at all, in
return. This trait will be selected and spread in the population,
until no one sounds the alarm anymore. So why do alarm calls
persist in many species? In some cases, the answer is to be found
in kin se lection. For example, yellow- bel ied marmots give alarm
calls, but not all of them do so equally. Most of the alarm calls
are given by mothers that have just had new pups. The pups, not
being as good as older individuals at spotting predators, are likely
to benefit substantially from the calls. Warning their pups is a
good investment for mothers, who don’t bother warning other
group members.11
A similar phenomenon might explain some alarm calls in Ara-
bian babblers: they live in groups of highly related individuals,
and the calls might boost the fitness of the caller by helping its
offspring, or the offspring of its siblings, survive.12 But that does
not explain why floaters— solitary babblers— also give out alarm
calls, even though they have no one to warn.
Surprising Communication Successes
The logic of the evolution of communication explains why it can
be so hard for individuals who have much in common— a mother
and her fetus—to communicate efficiently. It also explains how
communication emerges between individuals who seem to be
v i g il a n c e in c o mmuni c at i o n 23
locked in purely adversarial relationships. Even though the
existence and the extent of common incentives matter, what
matters even more is the possibility—or lack thereof—of
keeping signals honest and thus mostly beneficial for those
who receive them.
What common incentives do predator and prey have? Neither
wants to waste resources. If a prey is nearly certain to escape its
predator, they are both better off if the predator doesn’t attack
at all, and they can both save some energy. But prey can’t simply
send predators a signal meaning “You can’t catch me!” All prey
would have an incentive to send this signal, even if they were too
young, old, tired, hurt, or unprepared to escape the predator.
Predators, then, would have no reason to believe the signal. For
such a signal to function and to last, it should be disproportion-
ally likely to come from prey fit enough to escape. Other wise,
it is not evolutionarily stable, and so it will be selected out and
eventually dis appear (or never appear in the first place).
This might be what the alarm call of the Arabian babbler
achieves. By giving an alarm call, a babbler tel s the predator that
it has been spotted. Once it has been spotted, the predator’s
chances of launching a successful attack are low, as the babbler
can now seek cover. Many species, from lizards to kangaroo rats,
warn predators in this way.13 What keeps the signal honest, guar-
anteeing its evolutionary stability? Why don’t babblers emit
these calls at frequent intervals, just in case there happens to be
a predator around? One reason is that the calls don’t always deter
predators; they simply lower the odds of an attack. If the prey
has already been spotted by a predator, giving the call is worth-
while. But if the prey hasn’t been spotted, then it just made its
position known to any predators nearby and, since it doesn’t
know where these predators might be, its chances of escape are
low. As a result, prey have an incentive to give the calls only when
24 ch ap t er 2
they have actually spotted the predator, making the calls
credible.
Predator-deterrent signals have some intrinsic credibility, but
they can be made even more convincing by the prey orientating
toward the predator in a very vis i ble manner— something the
prey could not do without having already spotted the predator,
making the signal even more credible.14 For instance, Thomson’s
gazelles turn their rump toward a predator when they have spot-
ted one. The rump has a white patch, making it easier for predators
to get the signal.15 In this way, the gazelle can signal to its predator
that it has been spotted while still facing the other way in case the
predator decides that rump is too appetizing to pass up on.
Thomson’s gazelles not only show their rump to predators.
They also stot. Far from being useless, these high jumps also act
as a predator-deterrent signal. The gazelles are tel ing predators
that they are so fit that they would be sure to outrun them, so why
bother? Stotting is a reliable signal because only a fit gazelle can
stot enough, and high enough, to dissuade predators.
Stotting provides a good illustration of the kind of evidence
used to test an evolutionary hypothesis. How can we know
that the main function of stotting in Thomson’s gazelles is to
deter predator pursuit? First, we can rule out some alternative
hypotheses. Stotting does not increase the gazelles’ speed; in-
deed, they stop stotting when the predators get too close.16 Stot-
ting isn’t used to avoid obstacles, since gazelles usually stot even
though there’s nothing in the way. Stotting doesn’t simply signal
to the predators that they have been spotted, since the gazelles
rarely stot when they spot a cheetah. Cheetahs are ambush preda-
tors and thus don’t care about the gazelle’s ability to sustain a
long race.
Having ruled out these alternatives, what positive evidence
is there in favor of the hypothesis that stotting has the function
v i g il a n c e in c o mmuni c at i o n 25
of deterring predator pursuit? First, the gazelles stot in response
to the right predators: wild dogs, which are coursers. This makes
sense if they are advertising their ability to run fast for a long time.
Second, the gazelles stot more when they are fitter (in the wet
season) than when they are less fit (in the dry season). Third,
stotting works: wild dogs are less likely to chase gazelles that stot
more, and once the chase has begun, they are more likely to
switch toward gazelles that stot less.
How to Send Costly Signals for Free
Natu ral se lection has stumbled on impressively creative ways of
keeping communication honest, even in very adversarial rela-
tionships, by making it essentially impossible to send unreliable
signals. Babblers can’t coordinate their calls with those of unseen
predators. Only a fit gazelle can stot convincingly. Yet, human
beings do not seem to have any comparable way of demonstrat-
ing the reliability of the messages they send. With a few anec-
dotal exceptions— such as saying “I’m not mute,” which reliably
communicates that one isn’t mute— there are no intrinsic con-
straints on sending unreliable signals via verbal communication.
Unlike an unfit gazelle that just can’t stot well enough, a hack is
perfectly able to give you useless advice.
A commonly invoked solution to keep human communica-
tion stable is costly signaling: paying a cost to send a signal
edly explains many bizarre human be hav iors. Buying luxury
brands would be a costly signal of wealth and status.17 Con-
straining religious rituals— from frequent public prayer to
fasting— would be a costly signal of one’s commitment to a
religious group.18 Performing dangerous activities— from turtle
hunting among the Meriam hunter- gatherers to reckless driving
26 ch ap t er 2
among U.S. teens— would be costly signals of one’s strength
and competence.19
Costly signaling is often invoked but often misunderstood.
Intuitively, what matters for a costly signal to work is the cost paid
by those who send the reliable signal: it would be because some-
one pays more than a thousand dol ars to buy the latest iPhone
that owning this phone is a credible signal of wealth. In fact, what
matters is that, compared with reliable signalers, unreliable sig-
nalers incur a greater cost when sending the signal. In other
words, what matters isn’t the cost of buying the new iPhone per
se but the fact that spending so much money on a phone is cost-
lier for a poor person, who might have to skimp on necessities
to afford it, than for a rich person, for whom a thousand dol ars
might make very little difference.20
Given that what matters is a difference— between the costs
of sending a reliable and an unreliable signal— the absolute level
of the cost doesn’t matter. As a result, costly signaling can, coun-
terintuitively, make a signal reliable even if no cost is paid. As
long as unreliable signalers would pay a higher cost if they sent
signals, reliable signalers can send signals for free. The bower-
birds’ bowers illustrate this logic.
It is now well accepted that male bowerbirds build their bow-
ers to attract females. Indeed, better- decorated bowers succeed
in providing their builder more mating opportunities.21 Why
would females be attracted by fancy bowers? After all, these bow-
ers are of absolutely no practical use. Instead, Amotz Zahavi, a
biologist who did much to develop the theory of costly signaling,
suggested that bowerbirds indicate their value as mates by show-
ing they can pay the cost of building fancy bowers— which might
require taking risks, or going hungry while they forage for fancy