What is mirror recognition?

CSGnet Archive CSG1996
1

Rick,

Here's an example, a rather odd one, which might interest you. The
observations are as follows:

We're dealing with monkey's (macaques) and apes (chimps). The experimenter
surrepitiously places rouge on the primate's ear in the course of routine
handling. Then the experimenter arranges to get the primate in front of
the mirror. Chimpanzee's act like they recognize their mirror image; they
will scratch at the rouge on their own ear. Macaques do not do this. They
don't appear to have a clue that they're looking at their own image.

The standard conclusion is the apes have a more sophisticated sense of self
than monkeys. This is completely bogus. All that's been done is to put a
label -- "self-recognition" -- on the ape's behavior. But no explanation
is offered as to how it is that this mysterious "self" leads the ape to
scratch its rouged ear. That is taken to be self-evident, as though, when
the ape looks in the mirror, a neon sign proclaiming "self" lights up, the
ape sees and reads the sign, and concludes that it has rouge on its ear.

So, lets look at this in PCT terms. The form of the explanation would be
that the ape is able to control a perception which the monkey cannot. What
perception could that possibly be? What does the ape perceive that the
monkey does not? Both are in front of the mirror, both have very good
vision?

The explanation I came up with (but have never published, though it's in my
dissertation) is that the ape constructs more sophisticated intermodal
correlations. Imagine a monkey in front of a mirror. It sees the monkey in
the mirror. It moves around, and so does the monkey in the mirror. In
fact, there is a very precise correlation between the movements of these
two monkeys. But, the haptic/kinetic perceptual signals locate those
movements in one region of space while the visual perceptual signals locate
those movements in another region of space. In order for a monkey to
perceive the similarity between the sequences in its haptic/kinetic system
and the sequences in its visual system, the monkey has to be able to
overcome the different 3D spatial locations of those sequences. And, of
course, the monkey's nervous system has to have some way of locating its
own movements and the movements it sees in a common space, otherwise its
not going to do a very effective job of swing from vine to vine, etc. But
the same system which serves it so well in locomoting through trees, cannot
possibly afford to be so sloppy as to place the haptic/kinetic monkey in
the same 3D position as the visual monkey-in-the-mirror.

What if the ape had some tissue that allowed it to overcome this barrier,
that allowed it to perceive the similarity between the haptic/kinetic
sequence and the visual sequence even though they aren't in the same place?
Its efforts to regulate that perception might lead it to manually explore
its own body while observing the ape in the mirror. Hence, the ape is
acting in a way which the human observer interprets as evidence of
"self-knowledge." But this explanation, such as it is, has the advantage
of nowhere invoking this mysterious self.

I was led to this explanation by thinking about neural architecture.
Unfortunately that's a long story and it's time for dinner. For now, I
leave that as an exercise for the reader. Maybe I'll get through that
story by the time I have to take leave.

Another exercise for the reader: One of the various chimps who was raised
by humans, Vicki, once was sorting through a bunch of photos containing
pictures of humans and apes she knew. She placed all the humans in one
pile and all the apes in another. With one exception. She put her own
photo in the human stack. What's a reasonable PCT explanation of this? How
do you think Vicki would perform on the rouged-ear test? How's her
self-recogntion?

···

********************************************************
William L. Benzon 518.272.4733
161 2nd Street bbenzon@global2000.net
Troy, NY 12180 http://www.newsavanna.com/wlb/
USA
********************************************************
What color would you be if you didn't know what you was?
That's what color I am.
********************************************************

2

[From Bill Powers (961022.0530 MDT)]

Bill Benzon --

Chimpanzee's act like they recognize their mirror image; they
will scratch at the rouge on their own ear. Macaques do not do this. They
don't appear to have a clue that they're looking at their own image.

...

So, lets look at this in PCT terms. The form of the explanation would be
that the ape is able to control a perception which the monkey cannot. What
perception could that possibly be? What does the ape perceive that the
monkey does not? Both are in front of the mirror, both have very good
vision?

The explanation I came up with (but have never published, though it's in my
dissertation) is that the ape constructs more sophisticated intermodal
correlations.

Or what I would term "relationship" perceptions, a correlation being a
relationship between two or more variables (in this case, cross-modal). A
similar problem comes up when people try to draw something while watching
via a mirror. The mirror doesn't reverse the usual left-right relationships
between vision and kinesthesia, but it reverses the visual consequences of
making movements toward and away from the body -- if you write something in
the usual way on a horizontal surface it comes out looking upside down. So
most people have to do some reorganizing at the level of relationship
control to be able to write as skillfully as usual in this way. Did the
chimps at first have difficulty in moving their hands to the ear? If not, I
would suspect some kind of relevant prior experience, like seeing themselves
reflected in windows or shiny cage surfaces, or water.

What if the ape had some tissue that allowed it to overcome this barrier,
that allowed it to perceive the similarity between the haptic/kinetic
sequence and the visual sequence even though they aren't in the same place?

I don't think that "sequence" perception is the essential capability -- the
important aspect is a relationship. Also, I'm not sure whether "tissue" is
an adequate explanation; I don't think, for example, that skin "tissue"
would work. Since you don't like black-box explanations of perception, could
you explain how the brain tissue turns the visual and kinesthetic signals
into the perception of relationship (or correlation) to which you refer?

As to a mysterious "sense of self," I don't think this experiment tells us
anything about that, pro or con. I think that the sense of self involves a
higher level of perception, and is pretty hard to detect except by the
occupant of the system.

One of the various chimps who was raised
by humans, Vicki, once was sorting through a bunch of photos containing
pictures of humans and apes she knew. She placed all the humans in one
pile and all the apes in another. With one exception. She put her own
photo in the human stack. What's a reasonable PCT explanation of this? How
do you think Vicki would perform on the rouged-ear test? How's her
self-recogntion?

I don't know. Depends on what the basis was for the sorting. Obviously Vicki
wasn't doing the sorting on the basis of physical appearance, assuming her
form recognition was in working order. The data are too simple to support
any complex conclusions.

Best,

Bill P.

3

[From Rick Marken (961022.0730)]

Bill Benzon --

Chimpanzee's act like they recognize their mirror image; they
will scratch at the rouge on their own ear. Macaques do not do this. They
don't appear to have a clue that they're looking at their own image.

...

So, lets look at this in PCT terms.

Bill Powers (961022.0530 MDT) --

I don't think that "sequence" perception is the essential capability -- the
important aspect is a relationship.

It is also possible that the ape is controlling the relationship between two
visual configurations (the reflected hand and rouge configurations) with no
perception at all of any relationship between itself and what it is seeing.
The ape could have noticed that it can control this relationship just as
it notices that it can control any relationship (such as that between the
location of food and the location of its mouth). We might be interpreting
what the ape is doing in terms of our own higher level perceptions; if I were
doing it I would be controlling a perception of rouge on myself. But it's
possible that the ape is just controlling a perception of "red stuff on
a harry thing over there" with no awareness that the perception is of red
stuff on himself. One way to test this would be to simply show a picture of
a "rouged" ape (better, an experimenter dressed up as an ape) on a video
monitor. Set it up so that the "ape" in the monitor scratches at the rouge
when the ape who is watching scratches at himself in a corresponding
location. In this way the viewing ape can control what the ape on the monitor
is doing. You can also test whether the viewing ape is controlling a mirror
relationship between itself and the ape in the monitor; just set it up so
that, when the viewing ape scratches at his stomach, the monitor "ape"
scratches at the rouge on its head. See if the viewing ape still controls the
relationship on the screen. Mirror recognition may simply be "control of a
relationship that happens to exist on a reflecting surface".

Always testing for controlled variables.

Rick

4

[From Bruce Gregory (961022.1650)]

Rick Marken (961022.0730)

It is also possible that the ape is controlling the relationship between two
visual configurations (the reflected hand and rouge configurations) with no
perception at all of any relationship between itself and what it is seeing.
The ape could have noticed that it can control this relationship just as
it notices that it can control any relationship (such as that between the
location of food and the location of its mouth). We might be interpreting
what the ape is doing in terms of our own higher level perceptions; if I were
doing it I would be controlling a perception of rouge on myself.

Nice point. I have come to the conclusion that much of our
misunderstanding of others comes from attributing goals to them
that it would be proper to attribute to ourselves if we were the
person performing the behavior. This seems particularly true
where gender differences are involved. ("If I behaved that way I
would be indifferent to the other person's feelings...")

But it's
possible that the ape is just controlling a perception of "red stuff on
a harry thing over there" with no awareness that the perception is of red
stuff on himself. One way to test this would be to simply show a picture of
a "rouged" ape (better, an experimenter dressed up as an ape) on a video
monitor. Set it up so that the "ape" in the monitor scratches at the rouge
when the ape who is watching scratches at himself in a corresponding
location. In this way the viewing ape can control what the ape on the monitor
is doing. You can also test whether the viewing ape is controlling a mirror
relationship between itself and the ape in the monitor; just set it up so
that, when the viewing ape scratches at his stomach, the monitor "ape"
scratches at the rouge on its head. See if the viewing ape still controls the
relationship on the screen. Mirror recognition may simply be "control of a
relationship that happens to exist on a reflecting surface".

Vintage Marken!

Bruce

5

1. Looking at my schedule and my current to-do list I see massive
potential conflict. While, in the larger scheme of things, talking with you
folks on fundamental mind/brain issues is as important an activity as I can
do, in the short run .... So, this is where I have some slack and that
means that this post has to be my last (attempt at) a substantive post for
the time being. I may be able to get in some quick hits over the next week
or three, but no real thinking, etc.

2. On mirror recognition: I've looked back over my dissertation section
on this and on some of my sources. Unfortunately, I don't have a
photo-copy of one of my main sources, nor any substantial notes. The
article is: Gordon Gallup, "Chimpanzees: Self-Recognition," Science, 167:
86-87, 1971.

The problem is, not only to explain why chimpanzees do it (just whatever
"it" is), but why monkeys do not. As they are closely related species, this
is a tricky issue. But there are also other things which chimps do and
monkeys do not.

The most obvious example is language, whether signing or pressing panels
(with icons on them) on a board. Unfortunately neither signers & signal
boards or mirrors exist in the primate environment of evolutionary
adaptedness (of course, there are streams and ponds, but poking your hand
at the monkey in the stream gets very different results from poking your
hand at the monkey in the mirror).

However, there is also a difference between monkeys and apes on intermodal
learning. For example, it is easy to train a monkey to discriminate
between a circle and a square. If you then place a circular chip, a round
chip and the monkey's paw (monkey still attached) in a bag, you find that
the monkey cannot distinguish between the two, while people, of course,
can. My (2ndary) source says nothing about how apes perform on this, but
loosely implies they might be better than monkeys. Alas, I don't have a
primary source on hand.

A final piece of information. Chimps raised in social isolation cannot do
"mirror recognition."

Bill Powers (961022.0530 MDT) sez:

Or what I would term "relationship" perceptions, a correlation being a
relationship between two or more variables (in this case, cross-modal). A
similar problem comes up when people try to draw something while watching
via a mirror. The mirror doesn't reverse the usual left-right relationships
between vision and kinesthesia, but it reverses the visual consequences of
making movements toward and away from the body -- if you write something in
the usual way on a horizontal surface it comes out looking upside down. So
most people have to do some reorganizing at the level of relationship
control to be able to write as skillfully as usual in this way. Did the
chimps at first have difficulty in moving their hands to the ear? If not, I
would suspect some kind of relevant prior experience, like seeing themselves
reflected in windows or shiny cage surfaces, or water.

In one set of experiments the chimps were in a cage and the mirror was
outside. After several days the chimps started exploring their own body
while looking in the mirror. Monkeys did not.

Yes, relationships, but over what arguments and how constructed? The
"tissue" I had in mind was neural tissue in which to construct these
relationships (I should have been clearer about that). The visual sequence
is in one cortical region, the haptic sequence in another, and the motor
sequence in still another. there certainly are connections between these
regions; haptic feedback is important in guiding motion and some kind of
visual-motor connection would be necessary for imitation ("monkey see,
monkey do"). My sense is that to construct an explicit reltionship between
these three sequences requires more than connections between the regions.
It requires some tissue, either in a distinct region or scattered about in
these other regions, in which the patterns stored are patterns of
relationships between visual & motor and/or haptic sequences.

So, it it's relationships, then does that imply that chimpanzees can
construct relationships while monkeys cannot? Lets consider that
large-scale behavior survey I suggested last week. And let's imagine that
it turns out the way I expect it to, that all behaviors can be ordered such
that if a species exhibits the behavior(s) at position X on the list, it
also exhibits behaviors at positions 1 thorugh X-1; of course, we can
certainly have more than one behavior at a given position on this list.

The strongest possible result would be that chimps exhibit all the
relationship behaviors while monkeys exhibit none. A somewhat weaker
result would be that monkeys exhibit some (maybe even many) but not all
relationship level behaviors. If the result we get is this second one,
then we have at least two classes of relationships and we need to think
about that difference. Perhaps the difference is in how these two (or more)
classes of relationships are constructed.

One possibility that interests me is that chimps have localized
reorganiztion for relationships while monkeys do not. That is, here and
there scattered in the monkey's various memory boxes, relationships
occasionally get constructed. But only with much effort. This would imply
that, if one worked with a monkey for a long time, that monkey might begin
to display "mirror recognition." One doesn't have to work with a chimp to
get this; the chimp figures it out on its own.

As to a mysterious "sense of self," I don't think this experiment tells us
anything about that, pro or con. I think that the sense of self involves a
higher level of perception, and is pretty hard to detect except by the
occupant of the system.

Agreed, more or less. The experiment tells nothing about self. Toward the
end of the brain paper Hays and I have a few remarks about this "self,"
building on some information about self representation in the brain.
Beyond that, I've written a paper on the evolution of narrative and the
self (at my web site) which takes as a premise that expressive narratives
are, among other things, vehicles for the representation of "the self." In
that light, it does seem that, over the long haul, there is a definite
evolution of the self. The self perception/construction in a folk-tale is
considerably simpler than that in a novel. And it's not that folktale-only
people and novel people have much the same kind of self, but the novel
people perceive more of it. Rather, narratives are vehicles for creating
that self (seeing is being). So, the self of the narrative people is more
sophisticated than that of the folktale people.

One of the various chimps who was raised
by humans, Vicki, once was sorting through a bunch of photos containing
pictures of humans and apes she knew. She placed all the humans in one
pile and all the apes in another. With one exception. She put her own
photo in the human stack. What's a reasonable PCT explanation of this? How
do you think Vicki would perform on the rouged-ear test? How's her
self-recogntion?

I don't know. Depends on what the basis was for the sorting. Obviously Vicki
wasn't doing the sorting on the basis of physical appearance, assuming her
form recognition was in working order. The data are too simple to support
any complex conclusions.

There are other reports of human-raised chimps who seem to identify more
with humans than with chimps. I think we're looking at social perception
here. Vicki was raised among humans so those are the kinds of creatures
she interacts with.

3. And final comment on mirroring and imitation, this time in humans. Back
in college I took a course in developmental psych. The summer after that
course I was in the yard and one of the local 2-year olds (maybe 18 mo.)
wandered over. Since I'd just learned that kids like to mimic adults, I
thought I'd have some fun and turn the tables on the type by imitating him.
So, there we were face to face. He'd stick out his left arm, and I'd
imitate by sticking my left arm out. We went on like this for awhile when
it began to dawn on my that, not only was I imitating him, but
simultaneously, he was imitating me.

How could this be? He does something at t1. I do the same thing at t2.
He does something else at t3. I imitate it at t4. How could it be that, at
t3 he was imitating what I did at t2?

Of course, our actions weren't exact, etc. But... I realized that he didn't
understand left and right. So, he sticks out his left arm. I imitate him
by sticking out my left arm. From my point of view, we are doing the same
thing. From his point of view we are not. Imagine a line from his nose to
mine. His left arm is on one side of that line, while my left arm is on
the other side. So, as far as he was concerned, sticking his right arm out
is imitating me because it lines us up as though one of us were the mirror
image of the other.

later, much later,

Bill B

···

********************************************************
William L. Benzon 518.272.4733
161 2nd Street bbenzon@global2000.net
Troy, NY 12180 Account Suspended
USA
********************************************************
What color would you be if you didn't know what you was?
That's what color I am.
********************************************************