The trouble with language is its apparent altruism. Most speech,
except for commands and questions, appears to transfer potentially
useful information from speaker to listener. Speaking costs the
speaker time and energy, and brings information benefits to the
listener, so it looks altruistic. But, as we saw in the last chapter,
evolution tends to avoid altruistic behavior.
Fifty years ago, altruistic communication did not seem such a
problem. The animal behavior researcher Konrad Lorenz supposed
that communication was for the good of the species.
Animals could save their species lots of time and energy by
evolving signals that reveal their intentions and motivations,
especially in combat and courtship. This would reduce the deaths
from combat and the confusions of courtship. Ritualized threats
such as a dog's growling were supposed to convey accurate
information about the dog's level of aggression and willingness to
fight over a resource. If a growly dog meets a non-growly dog, the
non-growly dog should back down, saving the species a wasteful
dogfight. For several decades, the biologists' dogma was that
animal signaling meant communication, communication revealed
emotions and intentions, and communication evolved to make a
species work more efficiently.
The rise of selfish-gene thinking in the 1970s shattered this
idyllic view of animal signaling. Traits did not evolve for the good
of the species. In their seminal 1978 paper, Richard Dawkins and
John Krebs argued that animals should evolve to produce signals
only when signaling gives them a net fitness benefit that helps their
own genes replicate at the expense of other genes. Evolution cannot favor altruistic information-sharing any more than it can
favor altruistic food-sharing. Therefore, most animals' signals
must have evolved to manipulate the behavior of another animal
for the signaler's own benefit. Dogs growl because it was easier for
them to intimidate a rival than to fight. Smaller dogs could be
intimidated by deep growls because a deep growler is probably a
larger dog that would beat them in a fight anyway. Both the growl
and the growl-sensitive ears evolved for selfish reasons.
The modern theory of animal signaling grew from this
insight. Signals don't usually convey information about the
world, because signalers have so many reasons to lie about the
world. The theory suggests that animals usually evolve to ignore
the signals from other animals that may be attempting to
manipulate them. There are only a few exceptions. Predators
listen to signals from prey that reliably say "You can't catch
me," or "I'm poisonous." (Animals hiding from predators also
evolve camouflage, the purpose of which is to hide signals of
existence rather than to broadcast them.) Relatives listen to
signals from other relatives that reliably say "Watch out for that
predator!" Animals competing for a resource listen to signals
that reliably say "I could kill you." And animals looking for a
good mate listen to signals that say "I have good genes."
Basically, that's it. Except for the warning signals about poison
and predators, these signals are all fitness indicators. Any other
kind of signal that evolved in nature would probably be pure
manipulation, making the listener vulnerable to lies, sweet talk,
and propaganda.
The handicap principle can make fitness indicators reliable. It
can do so because the signal's cost is in the same currency—the
currency of biological fitness—as the signal's information. This
can work not only for fitness indicators that advertise good
condition to potential mates, but for signals of desperation that
advertise poor condition to relatives. For example, the handicap
principle may also account for the effectiveness of a baby bird's
gaping-mouth hunger display. Desperation signals also work with
the currency of fitness: the animal reliably shows how much a
desired resource would improve its fitness. Basically, fitness indicators advertise good condition and desperation indicators
advertise poor condition. Signals between unrelated animals
can convey information only about the signaler's own
condition, broadly construed. There are no credible models
showing that evolution can favor signals that carry any other
kind of information, as long as there are incentives for
deception.
This is a crippling problem for almost every existing theory of
language evolution, but the problem is not widely understood.
The handicap principle is not a magic wand that makes all communication
truthful just because a speaker has paid a fitness cost.
It cannot guarantee that a sentence conveys valid information. For
example, just because someone accepted the pain and risk of
infection necessary to get a tattoo does not make the tattoo's
message valid. It just implies that the tattooed person is stoical and
healthy.
Anthropologist Chris Knight has emphasized that human
language is especially vulnerable to deception because it
depends so much on "displaced reference"—referring to things
that are distant in time or space. To a person dying of thirst, we
can say, "There's a river over that hill." But displaced reference
is hard to verify. We might be lying about the river, and the
thirsty person might die if he goes over the hill expecting to
reach a river and finds a desert instead. In fact, there are no
theories of animal signaling in which reliable displaced
reference could evolve, given significant conflicts of interest
between signaler and receiver. Bee dances use displaced
reference to indicate the direction and distance of food, but the
bees are sisters from the same hive, so they have common
interests. Between our Pleistocene ancestors there were always
conflicts of interest, so it is very hard to see how reliable
displaced reference could have evolved. If displaced reference
was not reliable, listeners would not have bothered to listen, so
speakers would not have bothered to speak.
This brings us back to the altruism problem. At first it sounds
plausible to suggest that "language evolved to convey propositional
information from one mind to another." But that raises the question of why the speaker should altruistically give away
information to an evolutionary competitor. Truthful communication
is rare in nature because altruism is rare.there are three basic options for the hidden benefit:
kinship, reciprocity or sexual selection. The fitness benefits of
speaking must have come from giving useful information to a
relative, sustaining a mutually beneficial information-trading
relationship, or attracting a mate. I am sure all three were
important, and I am not going to claim that sexual choice was the
only selection pressure that shaped human language.
except for commands and questions, appears to transfer potentially
useful information from speaker to listener. Speaking costs the
speaker time and energy, and brings information benefits to the
listener, so it looks altruistic. But, as we saw in the last chapter,
evolution tends to avoid altruistic behavior.
Fifty years ago, altruistic communication did not seem such a
problem. The animal behavior researcher Konrad Lorenz supposed
that communication was for the good of the species.
Animals could save their species lots of time and energy by
evolving signals that reveal their intentions and motivations,
especially in combat and courtship. This would reduce the deaths
from combat and the confusions of courtship. Ritualized threats
such as a dog's growling were supposed to convey accurate
information about the dog's level of aggression and willingness to
fight over a resource. If a growly dog meets a non-growly dog, the
non-growly dog should back down, saving the species a wasteful
dogfight. For several decades, the biologists' dogma was that
animal signaling meant communication, communication revealed
emotions and intentions, and communication evolved to make a
species work more efficiently.
The rise of selfish-gene thinking in the 1970s shattered this
idyllic view of animal signaling. Traits did not evolve for the good
of the species. In their seminal 1978 paper, Richard Dawkins and
John Krebs argued that animals should evolve to produce signals
only when signaling gives them a net fitness benefit that helps their
own genes replicate at the expense of other genes. Evolution cannot favor altruistic information-sharing any more than it can
favor altruistic food-sharing. Therefore, most animals' signals
must have evolved to manipulate the behavior of another animal
for the signaler's own benefit. Dogs growl because it was easier for
them to intimidate a rival than to fight. Smaller dogs could be
intimidated by deep growls because a deep growler is probably a
larger dog that would beat them in a fight anyway. Both the growl
and the growl-sensitive ears evolved for selfish reasons.
The modern theory of animal signaling grew from this
insight. Signals don't usually convey information about the
world, because signalers have so many reasons to lie about the
world. The theory suggests that animals usually evolve to ignore
the signals from other animals that may be attempting to
manipulate them. There are only a few exceptions. Predators
listen to signals from prey that reliably say "You can't catch
me," or "I'm poisonous." (Animals hiding from predators also
evolve camouflage, the purpose of which is to hide signals of
existence rather than to broadcast them.) Relatives listen to
signals from other relatives that reliably say "Watch out for that
predator!" Animals competing for a resource listen to signals
that reliably say "I could kill you." And animals looking for a
good mate listen to signals that say "I have good genes."
Basically, that's it. Except for the warning signals about poison
and predators, these signals are all fitness indicators. Any other
kind of signal that evolved in nature would probably be pure
manipulation, making the listener vulnerable to lies, sweet talk,
and propaganda.
The handicap principle can make fitness indicators reliable. It
can do so because the signal's cost is in the same currency—the
currency of biological fitness—as the signal's information. This
can work not only for fitness indicators that advertise good
condition to potential mates, but for signals of desperation that
advertise poor condition to relatives. For example, the handicap
principle may also account for the effectiveness of a baby bird's
gaping-mouth hunger display. Desperation signals also work with
the currency of fitness: the animal reliably shows how much a
desired resource would improve its fitness. Basically, fitness indicators advertise good condition and desperation indicators
advertise poor condition. Signals between unrelated animals
can convey information only about the signaler's own
condition, broadly construed. There are no credible models
showing that evolution can favor signals that carry any other
kind of information, as long as there are incentives for
deception.
This is a crippling problem for almost every existing theory of
language evolution, but the problem is not widely understood.
The handicap principle is not a magic wand that makes all communication
truthful just because a speaker has paid a fitness cost.
It cannot guarantee that a sentence conveys valid information. For
example, just because someone accepted the pain and risk of
infection necessary to get a tattoo does not make the tattoo's
message valid. It just implies that the tattooed person is stoical and
healthy.
Anthropologist Chris Knight has emphasized that human
language is especially vulnerable to deception because it
depends so much on "displaced reference"—referring to things
that are distant in time or space. To a person dying of thirst, we
can say, "There's a river over that hill." But displaced reference
is hard to verify. We might be lying about the river, and the
thirsty person might die if he goes over the hill expecting to
reach a river and finds a desert instead. In fact, there are no
theories of animal signaling in which reliable displaced
reference could evolve, given significant conflicts of interest
between signaler and receiver. Bee dances use displaced
reference to indicate the direction and distance of food, but the
bees are sisters from the same hive, so they have common
interests. Between our Pleistocene ancestors there were always
conflicts of interest, so it is very hard to see how reliable
displaced reference could have evolved. If displaced reference
was not reliable, listeners would not have bothered to listen, so
speakers would not have bothered to speak.
This brings us back to the altruism problem. At first it sounds
plausible to suggest that "language evolved to convey propositional
information from one mind to another." But that raises the question of why the speaker should altruistically give away
information to an evolutionary competitor. Truthful communication
is rare in nature because altruism is rare.there are three basic options for the hidden benefit:
kinship, reciprocity or sexual selection. The fitness benefits of
speaking must have come from giving useful information to a
relative, sustaining a mutually beneficial information-trading
relationship, or attracting a mate. I am sure all three were
important, and I am not going to claim that sexual choice was the
only selection pressure that shaped human language.
No comments:
Post a Comment