Vous êtes sur la page 1sur 8

So, the next thing, I want to discuss

here, and the second phenomenon for today,


is Aperture Effects, which have to do,
not with the speed, well, it does
have something to do with the speed, but
it's primarily to do with the direction.
It's a set of phenomenon that,
I've shown you before, and
let me show you again,
just to remind you what these are.
So what you are going to see here,
is a line, you'll remember this
from previous presentations.
Line here moving from left to right,
and then the same line seen
occluded by, or
seen through better, better put.
The circular holes if you're
looking through a peephole.
So let's look at that.
Line moving from left to right.
Now the same line, the circle appears,
and the direction that you perceive
has gone from this to this.
And the question is why does that happen?
What, what's this telling us
about the perceived direction and
how it obviously differs from
the physical reality that's out there.
The physical reality of the same
line moving from left to right.
What's going on?
Well, the person who
the most to explain this
is a psychologist named Hans Wallach,
a very interesting man.
He was trained in Germany.
As you see he was born in the early
part of the 20th Century.
Trained in Germany.
He got his PHD in 1944 in Berlin.
And, he was Jewish so he had to emigrate.
Or he decided, very wisely,
to emigrate from Nazi Germany.
And came to,
the United States in [COUGH] 1935.
And came to Swarthmore College,
which is a small liberal arts college,
you may have heard of it, in Pennsylvania
and spent the rest of his career there.
His PhD thesis in Berlin.
He did many things over a very
illustrious career, but
his PhD thesis in Berlin
was about after effects.
And, let's look at what's actually,
happening, that's illustrated
in the diagram here.
Let's just, again the line moving from
left to right eh, through the aperture and

the perception of it is this way,


down and to
the right instead of the actual direction
of motion which is simply left to right.
And, you'll be aware,
if you think about this,
that a line that satisfies the apeture in,
in the sense of giving you.
The stimulus of a line moving
from left to right, or
a line moving through the aperture better
said, can be moving in any direction.
A line moving in this direction,
this direction.
In this direction.
In any direction,
can satisfy the aperture as long as
the ends don't appear within the,
the aperture.
If the ends appear in the aperture,
obviously.
The stimulus is quite different,
you no longer see the direction,
that you see when the ends are invisible,
when they're occluded, and when I use the
term satisfying the aperture, I mean that,
the lines ends don't appear and
a line can be moving in any of these
directions and, and
satisfy that condition.
Obviously they can not be moving
in the opposite direction,
that would not give this, but for
the hemispherical indicated here,
any direction is possible and
what Wallach did for his PhD thesis,
was to test and
describe the phenomenology and
of course in early 1930s
computers weren't available.
And he did this with assistants who
were actually moving metal rods
behind a circle or
some other form an aperture.
While he tested his perception of it and
tried to explain what was going on.
So let's look at this from
the perspective of another aperture.
Again, I showed this to you be, before.
But, while I tested many
different kinds of apertures and
probably the most famous of the apertures
is not a circle but a vertical aperture.
And we're looking at the same line.
We're going to be looking at the same
line moving from left to right, but
now occluded by a vertical aperture.
I'm moving left to right.
Vertical aperture.
And now you see the direction of the line

instead of moving left to right,


moving down.
Pretty much straight down.
Doesn't mean exactly straight down, and
that's important and
we'll come to that later.
But this is the,
the famous barber pole effect.
So if you, you know, remember what
a red and blue barber pole looks like,
if any of you still go to a barber shop,
and this is the, this is the and
Wallach was the first
really to describe that, in
in detail, and
Tried to explain it in different ways, but
he was I think basically
unsuccessful in that,
as other people have been in trying
to explain the aperture effect.
There are many papers, hundreds
literally on these phenomena, trying
to understand why we see direction of
motion in the strange way that we do when
the phenomenology that I've
just described is considered.
But suffice to say that there,
is no generally accepted
explanation of this, today.
So, let's discuss how we could,
analyze this in terms of human experience.
Not, not the speeds of moving stimuli,
but the directions of moving stimuli.
And the first step in, in making
this kind of analysis is determine
the directions that people actually see or
when a given aperture's applied.
So do some psychophysical testing.
And this is just a diagram of
the kind of test that's done.
Then the subjects are asked to
fixate on the center of the dot,
a line is moved through the aperture,
and they have an adjustable arrow,
that they can move until they think it's.
Representing, as closely
as they can judge,
the direction in which they're actually
perceiving the movement of the line.
So that's straightforward, and [COUGH]
the results of that are interesting.
So these are the reported
directions of mo,
motion in psychophysical testing for
a circular aperture.
And these circles here
just represent the degrees
around the circle but here in
relation to a direction that's normal
to the line moving through the aperture

which is taken to be zero degrees.


So, in this direction minus 30,
60, 90 degrees in this direction plus 30,
60, 90 degrees.
It's just a way of expressing
the psycho physical results.
Nothing complicated.
So what you see here is testing for
a line that's 30 degrees.
The orientation of
the line moving through.
The circular aperture is at 30 degrees,
here 35 degrees,
40 degrees, 45 degrees and
so on up to 60 degrees.
So testing different orientations and
what's the direction that
people see is allowing you through
the apertures changed and its, and
it's orientation as it traverses.
That space, and
what you see here is that, in every case,
the perceived direction is exactly
orthogonal, or close to orthogonal,
very close to normal,
to the orientation of the lines.
So, whatever the orientation of the line,
the direction that you see.
Is always orthogonal or
normal to that orientation.
So that's, that's the analysis
of the directions that people
are seeing as you vary the orientation
of the line in, in the aperture.
Now, again, the, what we're trying
to do here is see whether that.
Phenomenology, whether that
can be explained the way in
which we see the direction
of the line moving.
Can that be explained
in terms of experience?
Well, the technology is not out there
to generate that experience in the.
In a real world setting,
one has to again simulate it,
and in this case instead of moving dots,
which is what I showed you before,
here we're talking about populating
the thrust and the simulated space
of the real world, in this case that's
projecting onto the ersatz retina.
We populate it with bars moving
in many different directions at
many different speeds.
Again, within the range that
human beings are capable of
seeing by using speeds that
are outside that range.
But the difference between what I

showed you before, is not just,


rods versus dots, but also, an aperture,
in this case, a circular aperture.
So, what's the frequency of occurrence,
of projections through,
an aperture, on to the retina?
That's, our human experience,
that's what we've always seen.
And the next step then is to collect
the relevant empirical data simulated
in this case of lines projecting
onto the retina through an aperture.
Here's the results of that.
Again, kind of surprising and maybe we
need to think about this a little bit.
Oh, I'm sorry the lines moving
through aperture are different
orientations ranging from 60, or
through 30 rather to 60 degrees.
And these little override squiggles here,
are the observed probabilities of
occurrence of
lines of different lengths and
speeds moving through the aperture.
So in each case, the actual experience
that we human beings have when
we're looking at something through an
aperture is being defined by these little.
Probability ovoids that are telling
us what the frequency of occurance
of those lines projected onto the retina
have always been, in our human experience,
and the green arrows here,
in each case, represent the direction.
That determined earlier that
people actually see when they can
adjust the arrow and
say well what direction are they seeing.
That's always orthogonal to the line, so
the green lines are always orthogonal.
That's the human reports.
That's what we see.
So the question then is how does
that compare, how do these psycho
physical results, indicated by the green
arrow, compare with the frequency of
directions experienced over
evolutionary time, or individual life.
Indicated by the red arrows.
The red arrows, are the modes of these
little override probability distributions.
And you can see in each
case the red arrow and
the green arrow coincide nearly exactly.
Not always exactly.
I mean this is real data,
but the coincidence of the,
psychophysical result and
the absorb frequency recurrence.
In the simulated real world

with how rods project, or


how lines project onto the retina through
in circular aperture was remarkably good.
The up shot is that the mode of the
probability distribution of the projected
directions fits very closely
with what we actually perceive
in line with the idea,
that not only the speeds that we see, but
the directions that we see are determined
empirically by the evolution of trial and
error circuitry,
that allows our visual system to
judge direction based on the ranking
of experience that we've always seen.
Now, I, I need to explain to you wh,
why this is happening,
and it's not hard to see.
But why should these two things coincide?
Why should it be that the frequency
of occurrence of lines moving through
a circular aperture is very strongly
biased, here, let me go back here.
Remember that the red line,
the little ovoids that are determining or
showing us what the frequency of
occurrence of our experience has been in
terms of the frequency of occurrence
of lines moving through an apeture.
So, keep that in mind as we look at
this to try to understand well why,
why does that occur.
Why is it that it's much more frequent
to have lines that are moving in
this direction through the aperture.
Vastly more frequent then lines
moving in any other direction.
Lines do move in other directions but
this is.
The most frequent.
And why should that,
why should that be the case?
Well the bias arises because lines moving
orthogonally to their orientation would
satisfy the aperture more often than lines
moving in other, in any other direction.
So why is that?
Well, consider this red line,
which is the shortest line,
the shortest length of the line,
that can satisfy the aperture.
Again, if it's, any shorter than this,
you're going to see one end or the other.
And that's going to change the, stimulus.
You will not see the direction that we,
we see here.
And so that's the shortest line
that can satisfy the aperture.
Now consider any other line.
Any other line that's longer than this.

Because longer lines always


include this shorter line.
The frequency of occurrence
of shorter lines
satisfying the aperture is going to
be much greater than any other one,
simply because short lines
are included in longer lines.
So if you consider a line that's
longer than the red line here like
this blue line, well frequently for
any direction of its motion it's going
to satisfy the aperture because.
This red line is included in it.
But if it's moving in another direction,
then it's going to have to
be limited by its length in
the frequency of occurrence.
Satisfying the aperture, and that's simply
a way of saying that you're always going
to have a much greater prevalence
of lines that are moving orthogonal
through a circular aperture, in human
experience, than in any other direction.
So again, let's go back and look at this.
The ovaloids are the frequency
of occurance.
Of lines moving,
any line moving through the aperture.
Remember in our simulated environment,
there are lines of all possible lengths,
and moving in different directions.
So there's a straight forward geometrical
reason for this, and it's shown here.
Prevalent projected direction lines moving
in a circular aperture is always diagonal
to the line.
This is what humans will
sort of experience because
plausible explanation, for why we see
the direction of motion changing as we,
as we do in an aperture.
Where the same line in physical space.
Or the same rod in Wallach's
original experiments in this,
done in the 1930's, presents to us.
So, just to remind you of
the phenomenon of a vertical aperture.
Where the direction that we
perceive is moving downwards,
instead of to the right,
which is through a circle.
And they are many different
apertures that were all I tried and
that others have tried in
subsequent experiments.
And in, in [COUGH] all these
cases unless it's just
a demonstration of the same thing I just
showed you for the vertical aperture.

Exactly the same explanation applies,


the same simple geometrical bias applies.
So, let me sum up the main points
that I've tried to get across today.
First of all, motion perception
is just not well understood.
There's a lot to be said about motion
perception from the point of view of
the electo-physiology of recording from
experimental animal brains showing
that there are specific areas of the brain
which mentioned early on in the course.
And not going to say more about it here,
but there are areas that are And
specifically devoted, apparently,
to the analysis emotion, and
it's really not clear what
those areas are doing.
Taking the whole issue, from the point
of view of what can perception
tell us about how we need to think about
the physiology, and understanding.
The perception of motion,
generally the flash-lag effect,
and there's some related effects
that we didn't talk about.
Tells you, importantly that,
perceived speed can be
reasonably explained in terms
of accumulated human experience.
Same thing for aperture effects with
respect to perceived direction that
I've just told you about there is that.
The insight that was there
into perceived direction is,
again, that accumulated human
experience can explain via ranking
the phenomenology that we actually
see in terms of aperture effects.
And, the general result
of all of that is that,
if you consider the inverse problem, this
conflation of information on the retina,
that's the fundamental obstacle to vision,
in my view, and the theme of the course.
That an empirical explanation
based on accumulated experience is
probably the most feasible account
of the way in which we see motion.

Vous aimerez peut-être aussi