PSYCH 155/LING 155
syn
lab
UCI
COGNITIVE
SCIENCES
Psychology of Language
Prof. Jon Sprouse
Lecture 21: Animal Communication
1
Do animals have Language?
There is broad consensus among all biology-based scientists that animals have
communication: animals use various types of symbolic representations to
convey a message to conspecifics (and in rare cases, other species).
Today we are going to look at some of the more interesting communication
systems employed by various animals.
But the question of whether animals have language or not is tricky because it
depends on how you define “language”.
In conversations, most people use language to mean communication, in which
case the answer is undoubtedly yes.
Cognitive scientists on the other hand, use language to mean precisely the
system of representations and processes that humans have, in which case the
answer is undoubtedly no.
What we will see is that the communication systems of animals range from
relatively simple to relatively complex, but none reach the complexity of the
human communication system that we call language.
2
The bee “waggle” dance
Female scout bees go out and search for pollen and nectar.
After they find this “food source”, they come back to the hive to tell the other
bees where they found the food.
They communicate this location with a dance:
http://www.youtube.com/watch?v=-7ijI-g4jHg
3
Information contained in the dance
1. The direction of the food source
Bees dance on a vertical wall. Straight up represents the direction of the
sun. The angle of the waggle run represents the direction of the food
source in relation to the sun!
2. Distance of the food source
The length (in time) of the waggle run
represents the distance to the food source.
Different bee species have different ratios,
but a common figure is that 75ms of
waggling means 100 meters of flight.
3. Quantity of the food source
When a food source is better, a bee will shorten the time spent on the
round bits, thus spending relatively more time on the waggle run, and
also waggle more vigorously.
4
An interesting question: external or
internal reference?
Communication involves the use of symbolic relationships: a word stands for
its meaning, or a series of beeps in morse code stands for a letter.
An interesting question about symbolic relationships is whether the thing being
specified is internal to the communicator, or external (out in the world):
hunger = internal to the communicator
the dog = external to the communicator
Symbolic relationships that involve internal objects or states are in some sense
simpler than relationships involving external objects: think about how
shivering can signify that you are cold or afraid -- it doesn’t take any complex
“thought” to communicate this, it is simply a reflex of the body in response to
an internal state.
So it is an interesting question whether bees use symbolic relationships
involving external objects or not...
5
An interesting question: external or
internal reference?
In the case of the bee dance, various parts of the dance stand for pieces of
information necessary for finding the food:
1. The angle from vertical represents direction.
2. The length in time of the waggle run represents distance.
3. The energy expended dancing represents quality of the food source.
These feel external to us, because that is how we would describe a food source
(using external referents). But are they really?
When investigating the cognitive systems of another species (including
communication), it is crucial that you do not assume that behavior that
are similar to humans involve the same cognitive systems.
6
An interesting question: external or
internal reference?
In the case of the bee dance, various parts of the dance stand for pieces of
information necessary for finding the food and feel external -- Can we find a
way to interpret these as internal referents?
1. The angle from vertical represents direction.
Angle could represent the angle the bee turned, which is an
internal state.
2. The length in time of the waggle run represents distance.
Distance could represent the amount of energy the bee
expended, which is an internal state
Bees with weights on their legs overestimate distance in their
dances, suggesting that it really is about energy not distance.
3. The energy expended dancing represents quality of the food source.
The quality of the food could represent an internal state of
excitement about the food source (not the food itself).
7
Bird Calls
Bird calls are a basic example of symbolic representations of internal states.
Birds use calls to signify lots of different internal states:
1. I am taking off/landing
2. I exist
3. I have food
5. I am dominant/submissive
etc..
This is not to say that bird calls are not complex. They can be very complex.
For example the Black-capped Chickadee has a set of calls for inter-bird
conflict that is composed of 4 notes: A, B, C, D
The actual call takes the form A* B* C* D*, which means that any number of
A notes is followed by any number of B notes, which is followed by any number
of C notes, which is followed by any number of D notes.
What is interesting is that the number of each note conveys a message: the
number of A’s indicates the likelihood of attack, the number of D’s indicates the
likelihood of retreat.
8
Bird Songs
Bird songs are different from calls -- they tend to be much longer, involve
more notes, and exhibit more complex patterns
The function of bird song is related to territory-marking and mate-attraction -therefore males tend to be the singers and both females and males tend to be
the listeners.
song
sparrow
swamp
sparrow
http://www.biology.duke.edu/nowicki/
9
Bird Songs have complex
representations
The representation of sparrow songs has two parts: (i) the specific notes being
used, and (ii) the pattern of the notes.
song sparrow
swamp sparrow
We can see this by creating artificial songs that cross the two properties:
1. Song sparrow notes and Swamp sparrow patterns
2. Swamp sparrow notes and Song sparrow patterns
Female Song and Swamp sparrows do not respond to these aberrant songs.
10
Variability in Bird songs
Some species of birds have several songs in their repertoire:
Common yellow-throat: 1 song
Carolina wren: 2 - 10 songs
The marsh wren: > 100 songs
The northern mockingbird: unlimited!
11
Interesting question: How much of the
complex representation is learned?
For most birds, the songs that they sing are genetically specified.
This isn’t to say that they can’t learn variability. In fact, many birds
demonstrate regional dialects: small variations in the song that they learn from
their local neighbors.
However, in 3 (out of 27) orders of birds, the songs must be learned from
other birds, or they won’t be able to sing them properly! (songbirds, parrots,
and hummingbirds)
For example, white crowned sparrows go through the following learning stages:
0-35 days: no singing (but probably lots of learning)
25-40 days: subsong (like babies babbling)
35-80 days: “plastic” singing -- closer and closer
approximations of the full song
> 90 days: crystallization of the song
12
Alex the Parrot
Though parrots can be taught to mimic strings of human speech sounds, they
don’t really use those sounds as complex communication.
However, there is one example of a parrot being taught to use human-sounds
in a relatively complex way:
Alex was an African Grey Parrot that was
taught an extensive vocabulary of color
terms, number terms, shape terms, etc.
He demonstrated the ability to use those
terms to answer complex questions about the
world.
http://www.youtube.com/watch?v=R6KvPN_Wt8I
13
Vervet Monkeys
3 types of calls: eagles, leopards, snakes
http-//www.youtube.com/watch?v=hEzT-85gEdA
http-//www.youtube.com/watch?v=slGvI2y_W2c
Are they three distinct calls, or just three versions of “danger”?
Record a specific monkey’s leopard call, and play it back to a group of
other monkeys over and over again, even when no leopard is present.
Over time, the group will ignore that monkey’s leopard call, but will still
respond to the eagle and snake calls. This suggests that they are 3
distinct calls.
Is the reference external or internal? -- Probably external!
Vervets don’t make alarm calls when they are alone.
Submissive vervets make fewer alarm calls than dominant vervets.
14
Nim Chimpsky the Chimpanzee
Born in 1973
and raised by a
family in NYC as
part of a
research project
to teach
chimpanzees
sign language.
Learned about 125 signs over three years!
Major conclusions:
1. All (or nearly all) signs were performed when the referenced object was
visually or auditorily present in the environment. No discussion of non-present
objects except for requests for rewards (food, tickling, etc).
2. Though Nim could create several signs in a row, they tended to be repetitive
and without ordering constraints (no syntax):
GIVE ORANGE ME GIVE EAT ORANGE ME EAT ORANGE GIVE ME ORANGE
15
Chantek and Koko
Chantek the orangutan was taught signs just like
Nim, though actually in a more relaxed
environment.
It has been claimed that she knows about 150
signs, and will spontaneously start signing without
any prompting from humans.
But like Nim, she does not have any syntax associated with strings of signs.
Penny Patterson has claimed that Koko the gorilla
knows over 250 signs.
She also claims that Koko uses language just like
humans: swearing, making jokes, etc.
The problem is that the researchers have never
allowed anyone to see the data.
Similarly, they have tried to use Koko to inspire
donations to build a research facility in Maui...
16
(Matata and) Kanzi the bonobo
Matata was a bonobo (related to chimpanzees)
that Sue Savage-Rumbaugh tried to teach to
use a special keyboard. This keyboard used
colored keys with shapes to symbolize words
(instead of signs).
The idea was to teach them both the spoken
English words AND the keys for the words.
That way they could both hear and “speak”.
Matata was really bad at this. But she had a son named Kanzi who was too
young to be left alone. So Kanzi was present during Matata’s training.
Though Kanzi was never explicitly trained, when he became an adult (~2.5
years old), he demonstrated impressive abilities to use the special keyboard in
response to spoken English.
Kanzi can now use a keyboard with 256 words on it, and he can recognize
many more than 256 spoken English words!
http://www.youtube.com/watch?v=LxmbjLoUnhk
17
Kanzi and syntax
There is even some evidence that Kanzi
knows some word-order constraints.
For example, the order of actors in an
English sentence matters for the
interpretation of “who is doing what to
whom”:
Make the doggie bite the snake.
Make the snake bite the doggie.
http://www.youtube.com/watch?v=2Dhc2zePJFE
Kanzi can distinguish these two sentences of spoken English.
However, his syntax abilities seems to be restricted to these types of ordering
relationships.
For example, he has never learned that words like and mean that both actors
worked together:
The doggie and the snake bit the cat
18
Do animals have Language?
What we’ve learned is that the answer depends on the definition of “language”
No animals have the system of representations and processes that humans
have. And so far, we haven’t been able to teach that system to any other
animals.
But many species do have relatively complex communication systems that
involve complex representations:
Bees represent direction, distance/energy, and quality of food sources
Birds and monkeys have various types of calls that represent danger or
other “maintenance” functions, though they may all be “internal states”
Bird songs have complex representations and may require complex
learning, though songs don’t have complex meanings
Primates can be taught large inventories of signs (100-250), and can
combine them in meaningful ways, though not as complex as language
The much more interesting question is what representations do they have, and
ultimately, how do their brains process those representations!
19