Insight in Animals - PowerPoint PPT Presentation


PPT – Insight in Animals PowerPoint presentation | free to download - id: 714c6d-ODQxZ


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation

Insight in Animals


Insight in Animals Lecture 8 Psych 1090 – PowerPoint PPT presentation

Number of Views:47
Avg rating:3.0/5.0
Slides: 129
Provided by: imp60


Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Insight in Animals

Insight in Animals
Lecture 8 Psych 1090
Insight is complex
  • Refers to mental visualizations used to
    determine the best of alternate choices to
    solving a problem
  • Occurs not with respect to physical features
    that are obvious (e.g., nest sites), but features
    that must be imagined
  • That is, involves imagining the outcomes of many
    motor patterns or choices without ever having
    experienced them

And, once the outcomes are imagined, involves
choosing what to do, based on what is
theoretically imagined
Thus insight is the antithesis of costly
trial-and-error learning
Furthermore, insight must not be cued by other
events in the surroundings that might trigger
memories of related behavior
Insight may, however, involve integrating widely
differing bits of previously learned information
in an unconscious manner..
although some researchers argue that insight
requires consciousness
at least in the sense of conscious choice among
In any case, insight requires that a subject
  • project images forward in a progression
  • act based to some extent on general knowledge
    but not experience
  • understand the essential relationships of the
  • and have a specific goal in mind

In addition, demonstrating insight means ruling
  • not only any form of direct learning,
  • but also any form of prewired, innate behavior

Thus, one must examine problems that the animal
cannot have encountered
as a selective force in evolution
or through individual experience
The classic cases of insight, involving Köhlers
chimpanzees, of joining sticks and stacking boxes
to get food.
Has been criticized because animals
  • may have seen technicians performing some of
    these behavior patterns or
  • the number of choices available to the animals
    were too small (i.e., only one way sticks could
    fit together)

So, researchers want to examine insight in
controlled situations.
and, preferably, in animals that are less like
humans than apes to avoid over-interpretation of
but whose lifestyles would accommodate insightful
Heinrich chose to study insight in ravens because
these birds live in an environment that would
encourage such behavior.
One characterized by
  • unpredictability of resources and
  • the need to predict the behavior of others

Much of ravens hunting requires the ability to
determine where prey will go next.
e.g., small mammals, birds, reptiles that exhibit
escape maneuvers
So that they need to demonstrate at least a
learned capacity to predict the outcome of a
given situation
Many of their social interactions require them to
what either a conspecific or an allospecific
hunting partner (e.g., a wolf) will do next
Failure to act appropriately could have serious,
if not fatal, consequences
They steal food from other hunters
which again suggests that they must learn about
those hunters behavior patterns
And they cooperate among themselves at various
Ravens seem to exhibit intelligent, if not
insightful, behavior in nature.
One was observed to tackle a huge source of suet
by hacking a groove
making dozens of consecutive beak blows in a
precise line--
thus slicing off a piece it could carry
Others have been observed to throw stones at nest
And to wait to move until at least 5 of 6 hunters
had left a blind
if fewer than 5 entered, birds didnt move til
all had departed.
But these behavior patterns could have been
So Heinrich developed a task, string pulling to
obtain food,
that they would never come across in the wild
In particular, this task required a knowledge of
means-end behavior
which is not the same as simple associationif I
do X, Y happens
Understanding means-ends behavior implies the
ability to see a novel situation
Recognize the goal and how to achieve the goal
based on the qualities of the situation
and the qualities of the elements that constitute
the situation
Not all species succeed
Studies with some songbirds showed that they were
actually trained on the task in a stepwise manner
And for some species, previous experience turns
out to be important
Ravens dont seem to deal with situations in
which they have to, for example, pull vines to
get berries
BUTthey can grasp objects with their feet and
so string-pulling would be a possible task to
The idea was to determine if
  • meat was suspended low enough that they couldnt
    simply reach down and pull it up,
  • or break off chunks and fly away
  • could they reach down, pull up string, hold onto
    string w/ their feet, release their beak, and
    repeat the maneuver as many times as necessary to
    get the meat?

This behavior meant integrating a large number of
different motions
in a very precise order
to solve a problem never previously encountered
and to do it without trial-and-error learning
Too, if the birds succeeded on this task,
it could be made more complex to test related
and possibly insightfulabilities
The experiment was not simple
Wild birds flew away from meat suspended on the
Adult ravens tend to be neophobic, avoiding any
situation that has an unfamiliar component
which makes testing in novel circumstances almost
Wild-caught birds that had lived in an aviary for
a year werent much better
Only one of 14 aviary birds approached and pulled
up the meat in lt1 minute
Only three birds followed the firsts lead over
the course of a few days
In a second test, with a wild-caught, caged
group, 3 of 13 performed the behavior correctly
In a bunch of 4 month old juveniles, not yet
neophobic, 4 attempted the task but failed..
But 2 yrs later, 3 of these 4 birds succeeded
w/out intervening trials
So, Heinrich decided to test some hand-reared
That way, he could make sure what experiences the
birds had already had
These birds were 1.5 yrs old and into the age
where neophobia could be a problem
He used birds that had never seen string before,
but were used to people, so having experimenter
observe them would not affect the experiments
and were also used to small changes in their
environments, plus probably things like hoses
One bird pecked at the string and actually tried
yanking the string
But didnt get the idea of raising the string
Then jumped from below and managed to tear off
some tiny bit of meat
Other birds tried to fly up and get the meat from
below, but could not tear off the hard, dried
food that was used
One bird approached the string from the perch,
pecked at it and yanked once, then left the site
Then, 6 hrs laterthis bird completed the entire
string-pulling maneuver!
Heres a clip of one of my birds performing the
task, so you can remember what is involved
Clearly, that raven demonstrated insight.
It had observed the situation, gone away, and, in
the 6 hr interim,
somehow figured out what it would have to do to
And without further experimentation, except in
terms of mental representation
Four of the other five ravens in the group
succeeded in the same way
They had seen the first bird succeed,
and one copied the technique, of pulling string
along the perch
so they might be imitating or emulating that
birds behavior
Well talk about the difference in imitation and
emulation in detail in another lecture
But emulation is copying the goal without
necessarily copying the action
Imitation is copying both
But what level of insight was shown?
Did the birds actually understand the physics of
the situation?
More experiments were needed to test out the
details of what the birds did or didnt know
Interestingly, crows didnt succeed on these
but, as well see, two types of parrotsGreys and
making one wonder about exactly what cognitive
skills are involved
In any case, Heinrich kept testing
For example, if the meat at the bottom of the
string was very large
a sheeps head that the birds already knew they
couldnt lift it from prior experience.
would they still attempt the task or would they
go for the biggest bit?
No bird chose the 2kg portion all chose the
small piece
Of course, because the birds had had experience
with the sheep headnot salamion the ground,
maybe they expected more sheep head later without
having to work at it
And simply chose on the basis of expectation and
ease of access
The issue with all these studies is that one can
likely never totally rule out alternative
but must keep trying to do so with additional
Of interest, however, was that birds seemed to
expect that strings held meat.
Strings always did hold meat and the birds had no
reason to expect otherwise
Again, what did they know about the situation?
So, when given a choice between a string with a
rock and one with meat,
they would make one or two tugs before switching
to the meat-laden string
They never persisted with the wrong string,
but still made 10-20 errors before learning to
look down first!
Were they really being stupid?
Or just didnt bother because the cost of making
an error was so low?
Only way to tell was to make the task a bit more
Now that birds had learned to look, could they go
to a three-choice test where one string had
The birds had never had seen a new colored string
Most birds eventually figured out that they had
to go to the new string.
they made fewer errors than they had made with
the stone versus meat task.
but not all looked down at first
Thus there was some aspect of learning, rather
than only insight, involved in the behavior.
But now that the birds had learned to look, could
they understand causality and connection?
What if the experimenter crossed the strings
having the meat and the stone?
Were birds just pulling on string above meat, or
on string to which meat was attached?
Results depended upon how far apart the objects
were displaced
When items were displaced only a few cm,
most of the birds chose the correct string
But when the distance was increased to about 40
Three birds failedon over 79 trials!
Only one bird got it right, 17 times out of 21.
Which shows that the ability to understand the
situation is inherently possible in ravens
although not at all widespread
What would happen if the birds now saw a new
Would they be confused because it never had meat?
Nopethey did better, 32 of 33 trials
correct.these were different birds from the
earlier task w/ colored strings
Obviously, the different colors helped them to
distinguish which string was connected to the meat
But there were still aspects of the study that
needed to be clarified
For example Did the birds understand the
connection between the meat and the string?
Ravens generally take meat they are given to eat
somewhere else.
If the birds were shooed off after getting the
would they drop it or try to fly off with it?
That is, did they realize that the meat was
connected to the perch via the string?
None of the ravens that had pulled up the meat
tried to fly off with it!
Butone raven that never pulled meat, and several
crows that never figured out the task
ALL tried to fly off with meat that they were
given that was attached to a string.
Thus most of the ravens had understood the
string-meat relation
they didnt exactly learn it, although their
experience would have helped
but it clearly was not innate
But, if Heinrich just attached string to some
meat, and the birds did not have to pull it up.
would they act as though meatstringattachment?
Most birds immediately flew off with the meat,
realizing that there was no overt connection
Now, several problems with these experiments
still needed to be addressed
  • fear of strings
  • dominant birds keeping the subordinates from
    attempting the task
  • observational learning

So, now a new group of birds were given strings
in the aviary from the time they were put in the
Birds were tested individually, separated by an
opaque partition
so they wouldnt be disturbed or see one another
5 of 6 birds figured out the behavior in just a
few minutes
None of the birds did the task absolutely
All of them tested out simpler techniques that
allowed them to learn something about the
string-meat connection
But birds did not persist with these techniques
in a trial-and-error fashion
They learned something about the system then
applied that knowledge directly
solving the task in a very different manner than
they had used in those initial attempts
and using both loop and stretch techniques
Now, would the birds still be correct if they had
to do a different task involving meat and string?
This time, they had to pull down so that the
string and meat were pulled over a wire
Two groups of birds were used.
One group, who had been part of the earlier tests
and one bird that was totally naïve
The experienced birds all succeeded, whereas the
naïve bird failed..
This task was particularly important.
The naïve bird would have had to deal with a
novel task
one for which no memory could have served to
provide a mental representation for solution
whereas experienced birds had already learned
that pulling on strings led to getting meat.
But for the experienced birds, the task would be
Thus the experienced birds showed intelligence
in adapting what they had learned to a new
but maybe not insight
The failure of the naïve bird could either have
been specific to that bird
who wouldnt do the earlier task
Overall, the experiment showed that insight is
based on something learned,
even if transfer of that information is necessary
So Heinrich designed another experiment to
determine what was necessary for insightful
behavior to arise
this one involved knowing more about size and
weight of prey items
as well as the string-pulling behavior
If the birds had a good idea of the behavior and
what it required
they would immediately adjust their actions
if the weight of the meat were heavier and the
string slipped more.
And all birds did just that.
often with no change in the time it took to get
the meat
despite the need to hold the string much more
PBS ravens
So.lets go back and see how the ravens actions
compared with alternatives to insight.
  • innate programming
  • learning
  • random chance

Programming cannot account for behavior that
  • is not encountered in the wild
  • appears without prolonged experimentation
  • differs in detail among individual birds
  • is updated and altered in progress in accordance
    with different goals

Not much of the data is consistent with learning,
The birds appeared to have anticipated at least
some consequences of the behavior
before overtly executing their actions
Even if their first steps onto the string were
just random chance.
The birds behavior changed markedly within a
short time after that initial step.
Whether the insight occurred when some birds went
away and came back,
or when others started to examine the string is
but no bird engaged in prolonged trial-and-error
learning of the behavior
Before we look at dogs, what about other
supposedly intelligent birds?
As mentioned briefly, crows failed to do this
And in my Grey parrots, only those that have not
had significant language training succeeded.
Language-trained birds looked at the nut, then at
the trainer, and requested that the trainer
provide the nut
As we saw before, Arthur (and Kyo before him) had
no trouble picking up the nut without any prior
But Alex and Griffin, who had considerable
training in requesting items and having trainers
would not do the task
Note that Arthur, somewhat like Heinrichs ravens
differentiated a more favored nut from the less
favored chalk in terms of what he chose first
but didnt need multiple trials
Whether the language-training has caused my
birds brains to develop differently is unclear.
That is, might the communications centers of my
birds have developed to the detriment of their
manipulative behavior?
Only slight evidence to support this idea
When Griffin was developing object permanence, he
briefly plateaued on one task while he fledged.
But such behavior might reflect more about where
attention is being directed at a given time
than about overall brain development.
That my birds have learned to manipulate trainers
to get their needs met is quite likely.
to the extent that they dont see the task as one
they must solve on their own
The point is that not all species and not all
individuals of a species exhibit insight
But the reasons for the various behavior patterns
are more complex
than what could be seen as levels of cognition.
And sometimes experience can have a negative
effect on a particular behavior pattern of
Work with other avian species has not shown
behavior that is clearly insightful
For example, tits opening milk bottles to get
cream (when milk was not homogenized and home
was more a case of happenstance.
Tits peel bark back to get insects and larvae,
and the action on the bottle is the same..
Birds could have been attracted to the bottle in
the first place because of some spillage or the
and this behavior is a single, simple action,
rather that a pattern to be developed
What about another parrot species that is
supposedly extremely intelligent?
The New Zealand kea, that can destroy a backpack
in about 20 seconds
or quickly strip a car of its rubber gaskets
The kea is, in the wild, an extractive forager,
with much probing and prying as well as ripping
apart of items
has a complicated social system with a dominance
hierarchy not unlike that of apes
in sum, another potential subject
With the exception of a fledgling, who simply
seemed uncoordinated,
all six of Hubers keas (Animal Behaviour,
Werenich and Huber, 2006) solved the initial
(simple) string task on their first trials
without having seen anything like it beforehand
In a test with unbaited versus baited
stringsbut of different colors (which helped the
Five of seven subjects chose correctly on the
first trial
and the other two succeeded on the second trial
The keas were not quite so good with the crossed
string trialsbut not bad
Only one bird chose the correct string in the
first trial,
but that bird made only 2 more errors on the next
29 trials
Three other keas learned fairly quickly after
they failed on their first trial
making only 5 or 6 errors each on their following
29 trials
Two other birds were total ditzes, never getting
And one bird figuring it out after about 15
which would suggest learning
So individual differences are quite striking in
this group
Even with identical background
tho social structure not reported
Note that the researchers re- did the trials with
the same color strings
The birds had a much harder time, just like the
and this was after they had succeeded with
different color strings
Huber also tried a slanted string session.
To test if angling the strings mattered
Only one bird failed the test on the first trial,
But some failed after succeeding initially
which could suggest either playfulness or
given first trial success
The keas were also given the overload test
Here a stone was slathered in butter
Three birds failed on their first trials, and
others succeeded
But the birds that failed learned quickly
Not insight, but intelligence
What was very surprising was that when Huber gave
his birds an extremely long string
Such that they could just fly to the ground and
extract the reward that way
Only one bird changed her behavior!
One would think that the birds would figure out
that something had changed to make the task easier
But habit seemed to win out over insight
which is an important issue
Interestingly, in the wild, Johnston found that
only a small percentage of keas succeeded
Given their dominance hierarchy and what we saw
of social effects on the behavior in our Greys
lab versus wild may be important for social
contexts, too
Now, we arent going to go back over the
insightful behavior of the New Caledonian crows
Im just going to remind you how Betty altered
her behavior in novel ways depending on the task
she was given
Sometimes in unexpected ways
So, what about the dog paper?
If you took Blumbergs course in the Extension
or have been reading the work of the Hungarian
and German researchers
youd expect dogs to do just fine
Dogs, for example, can follow a human point to
find a hidden food
whereas apes need tons of training to learn that
And, as hunters, dogs should have some idea of
cause-effect actions
From the literature review, it appears that dogs
can be trained to do a string-pulling task
but that none of the previously tested subjects
performed the test via insight
but sample sizes were small, and thus there was a
need to re-test
One or two strings of different colors were used
Im not entirely sure that the dogs could detect
the different colors as well as the experimenters
but dogs do have color cones, and red is
supposedly very distinct
so lets assume that the color of the strings was
a non-issue
The dogs had to learn to use the straight
stringwhich in and of itself can be seen as
important or not
given that the birds did not need any training at
But dogs didnt have problems with long string
after learning about short string
Suggesting that they could at least transfer
easily within similar conditions
But most of the dogs didnt transfer easily to
the diagonal string
pawing at the edge of the box with food
Of course, some dogs had just used their tongues
would be interesting to see how the data would
look if those dogs were excluded from the
diagonal trials
But that was not done
One might have expected that the dogs had learned
to use the string and should have transferred
but maybe for the one set of dogs there was some
would just have been interesting to see if that
was the case
With two other sets of dogs, the experimenters
trained one set with one string connected to food
in two different locations
and the other set with two strings, one baited
and one not
to test if the conditions were important
Then dogs had long strings baited or not baited
and had to choose
sometimes the strings were far apart and other
times fairly close together
which could make a difference
Performance was better in near than far
Possibly comparison was easier?
In any case, the dogs could just paw at the place
closest to the food
So the experimenters had to really test if the
animals had any idea that the string was
connected to the food
and what that meant
So the dogs had two possible types of tests
One test would be with the odd string near the
The other would be consistent
If the dogs understood what they were doing
they would preferentially choose the string
attached to the food
and ignore what was just close in terms of
Now remember, however, that these dogs had not
been involved in the other tasks
But were separately trained on string pulling
tasks involving diagonal strings
that were always close or connected to food
The dogs learned the training situation fairly
But were pretty much at chance when the tests
involved strings that forced them to choose the
one connected, rather than close to, the food
Then a new set of dogs were given a crossed
string test
to see if they could deduce, after learning to
pull short strings connected to food
to switch to the strings at cross purposes to
what had been learned
But they couldnt switch
even over a number of trials
What could have happened if they were trained
from the start with sharply angled ones(not thru
And tested with the crossed?
We really dont know, and that is one
experimental situation that needs to be tested
And, might a dog do better with a token that it
could trade
Like Boysens chimpanzees?
So, we see some aspects of insight and learning
work together to demonstrate cognitive processing
in animals
but much is left to be discovered
and more species need to be tested
And, we still do not know what neuronal
mechanisms are involved in insight.
Clearly, insight must have evolved because it
supports goal-directed behavior
A subject that has memory and can form
representations can assign relative value to
proposed actions
The subject then is seen to be acting
But many humans, when queried as to how they
solved a problem via insight,
are at a total loss to explain the procedure
Such individuals consider insight to be some kind
of aha experience.
Yet the behavior is based on integration
of previous knowledge applied to a novel situation
But what is insight?
It clearly involves knowledge of ones
environment and actions in which one can engage.
It appears to be an emergent property that is not
widespread throughout the animal kingdom.
So, insight seems to be a mixture of
  • learned, stored information
  • mental representation of the problem
  • some realization, conscious or not, of what the
    solution should be
  • conscious choice to act

Clearly, the ravens and parrots exhibit behavior
that is complex and intelligent.
And clearly that has ramifications in the ravens
daily life..
For example, ravens feed with wolves at
carcasses, and could easily get bitten.
Even one bite could be fatal
But a raven that can project a wolfs movements
in relation to its own can be safe
What does this have to do with the parrots
Maybe as a prey animal, it needs to track the
or maybe parrots are just very good generalists
In sum, theres no real way to predict how a
particular species is going to react on a task
which is why comparative testing is so very, very