Perception

1
Chapter 7

• Perception

2
(No Transcript)
3
Sensation Vs. Perception, Again!
4
Brain Regions Visual Perception
Primary visual cortex is made up of a large
number of modules which contain a large number
of nerve cells that all respond to different
aspects of the same part of the retina termed
the visual field of those nerve cells. The
retina is not evenly represented but, instead,
more primary cortex is devoted to images at or
near the fovea. Some nerve cells in a module
respond only to lines of certain orientations,
others respond only to motion, others to colour,
etc. Thus, primary cortex codes the basic
features of the image it receives.
5
Brain Regions Visual Perception - 2
Secondary visual cortex regions (i.e.,
association cortex) is responsible for higher
level visual processes as revealed by various
types of brain injury Damage to primary visual
cortex - often results in blind spots but no
problems with object recognition. Damage to one
part of association cortex can lead to an
inability to see colour altogether, a problem
termed achromatopsia. Damage to a slightly
different part of visual association cortex can
result in an inability to perceive motion.
6
Visual Agnosia and Prosopagnosia
Perhaps the most interesting deficits occur when
the parietal region of visual association cortex
is damaged. Sometimes damage here leads to an
inability to identify objects despite normal
visual acuity - visual agnosia. Other times, the
damage results in an inability to recognize
faces, even those of very familiar people -
prosopagnosia. These inabilities can occur
despite a lack of problems with other complex
visual tasks such as reading.
7
Basic Issues - Figure vs. Ground
Figure vs. Ground One of the most basic issues
in visual perception concerns how we look at some
scene of an image and figure out what is figure
(the object of interest) and what is ground (the
context the figure occurs in). Edges and
countours are usually critical in this respect
and usually provide good information
8
Figure vs. Ground - Ambiguous Images
9
Basic Issues - Perspective
An issue related to figure/ground is the
following. Sometimes visual scenes are somewhat
ambiguous, and can be scene in different ways
what are the cues that allow us to see
one perspective over another? Is the picture on
the left a picture of a young stylish woman, or
of an old woman?
10
Other Reversible Figures
The classic reversible Necker cude and some more
stylish versions
11
Gestalt Laws of Grouping
According to a group of German Psychologists
called Gestalt Psychologists, the primary purpose
of the visual system is the recognition of
objects from basic visual elements. The objects
are seen as more than a sum of the parts, and the
critical problem facing the visual system is how
to group the elements to form objects. Several
principles, or laws, are used by the visual
system to do this grouping
12
Demonstration of the Importance of Objects over
Elements
When elements are arranged in groups that define
an object, we tend to see the object and not the
elements. This object-superiority can be
demonstrated in Stroop-like experiments that use
stimuli such as
FFFFFFF FF FFFF FF FFFFFFF
EEEEEE EE EEEE EE EEEEEE
Global interferes substantially with local
decisions, but there is much less interference
of local on global decisions
vs.
13
Law of Proximity
Things that are relatively close to one another
tend to be grouped together
14
The Law of Similarity
Items that look similar will be seen as parts of
the same form
15
The Law of Good Continuation
The tendency to perceive unseen parts of a
patterns as continuing in a predictable and
simple manner.
16
The Law of Closure
Often an object is partly occluded by other
objects in our environment, and the visual
system must fill in the missing information
17
A Related Phenomenon - Illusory Contours
18
The Law of Common Fate
Finally, elements of visual perception that move
together are seen as forming a common
object. This law is best imagined in terms of
those animals you see on nature shows that seem
to perfectly blend into their background, until
they move. Then suddenly they appear
visible. As an example though ...
19
Common Fate Example - 1
20
Common Fate Example - 2
21
Common Fate Example - 3
22
Common Fate Example - 4
23
Perception of Form - Summary
Thus, a number of laws help us to perceive form
that is, to figure out what the objects are, and
how to interpret them despite actual
sensations. Once again, these laws are The
law of proximity The law of similarity The law
of good continuation The law of closure,
and The law of common fate
24
Figuring out what the objects are
The Gestalt principles help us to understand how
we figure out what the objects are, and how to
interpret them. However, they do not explain how
we figure out what an object is once we realize
it is an object.
25
Templates and Prototypes
One idea about how we might recognize objects
centers around the notion of templates the
reverse cookie-cutter idea. Such templates seem
unlikely given the extremely huge number of them
we would need to recognize all the objects we
know from all the orientations we know them
in. However, a fuzzy template idea called
prototypes may work. The notion here is that we
might have some idea about what a typical version
of some object might look like, then we compare
experienced objects to these prototypes and
except the best match.
26
Evidence Supporting Prototypes
If subjects are asked to categorize items as
being birds or not, they can correctly categorize
more typical birds faster than they can
categorize less typical birds.
Imagine a penguin in this box (or a picture of a
penguin at least)
Slowest RT Middle RT
Fastest RT
27
Recognition via Distinctive Features
Another view is that we recognize objects via
distinctive features that define those
objects. For example, consider these examples of
the letter Z, what do they have in common?
Z Z Z Z
Z Z The answer, two
horizontal lines and one diagonal line. Perhaps
it is the presence of these features that define
an object as being a Z
28
Evidence Supporting Distinctive Features
Lets play spot the Z!
29
Evidence Supporting Distinctive Features
Lets play spot the Z!
OBCCGDOOPDDGQQCCPOCGD OPRGPOCBGQRQSSUOPCSRUP QCDBP
OSCURPOPPCDBZPODQ POQSGOPQCBBCGPOQDUOPQ OPQDCBGSOS
PQSRCBDOPQSC
30
Evidence Supporting Distinctive Features
Lets play spot the Z!
KLEFIXKNMLMVXWIYLKMNX IKLWNMVXAILKHNMVTEFILM IMKLN
VXWAVNMKLIYZFENM IMKLNHXVEYIFKLMNVWTYIL XVNMKLIYWT
NMLKMFWENM
31
Distinctive Features and Real World Objects
Features seem like a natural way to think about
how we identify letters, but what about real
world objects? Beiderman (1987, 1990) suggested
that a similar logic could be applied to real
world objects, except they need to be thought of
as being composed of geons instead
features. What the heck are geons? - basic three
dimensional shapes
Steve, show figure 7.17 at this point
32
Help from Context
In addition to the bottom-up recognition
achieved by the analysis of features, context
also provides a top-down way to bias the system
for perceiving some items over others. I have
already shown you several examples of this
including the one to the right here. Studies
with tachistoscopes show that this bias works two
ways as in the bread versus mailbox example of
Figure 7.22
33
Perception of Objects - Summary
Thus, a number of things may be crucial in our
ability to figure out what things are,
including The use of prototypes Analysis of
features Contextual Support
34
Answer to E-mail Question
Someone e-mailed me a question asking what it
means when someones eye jitters apparently
some cultures have some bizarre theories about
this.
As the following demo shows, all our eyes jitter
a certain amount all the time, we just dont
tend to notice it. It is possible for amount of
jitter to vary across people or time.
35
Spatial Information
OK, so now we figured out what sensations are
objects, and what those objects are the next
problem is figuring out where the objects are in
space. In fact, after leaving the primary visual
cortex, visual information seems to travel along
two fairly-distinct pathways, one pathway is
devoted to figuring out what things are, the
other is devoted to figuring out where they
are. There are many cases in the neuropsychology
literature of patients that can perform tasks
based on one of these sources of information, but
cannot do tasks based on the other.
36
Depth Perception
In order for us to figure out where an objects
is, we need some way of judging depth in our
visual environment. In turns out that we use a
fairly large range of cues to help in our
perception of depth, and the fall under two
general classes. Some depth information can only
be obtained when both eyes are viewing the world
these types of information are termed binocular
cues to depth. In contrast, monocular cues can
be obtained using only one eye.
37
Binocular Cues - Convergence
Because the two eyes converge on an object when
we are viewing it, the brain can use the angle of
convergence as a cue to how far away that object
is. For example
The larger the angle, the nearer the object
38
Binocular Cues - Retinal Disparity
Whenever we are not focusing on an object, the
image of that object falls on different points of
the two retinas. The amount of disparity
(difference) between the two retinal images can
be used as a cue for distance. Try holding up
two fingers one in front of the other. Focus
on the front one (you should now see two images
of the back one). Now move the back one away
from, then back towards you, while still
focusing on the front one. What happens to the
two images you see as the back finger moves?
39
Monocular Cues - Interposition
When one object partially occludes our view of a
second object, we assume that the first object is
closer to us that the second. For example
Vs.
40
Monocular Cues - Perspective
Things appear to get smaller as they recede into
the distance even though we know they are not
actually getting smaller. Given this, if the
general size of some object in the scene is
known, the size of the retinal image cast by that
object can be used to judge its distance from
us. This can sometimes lead to neat illusions
such as the one to the right.
41
Monocular Cues - Shading
We live in a world where our major light sources
tend to come from above. As a result, shading
has come to be another source of depth
information such that objects that are
shaded lighter on the top are seen as sticking
out towards us.
Steve, show Figure 7.32, and turn it around.
42
Monocular Cues - Texture, Haze Horizons
As illustrated in Figure 7.30, closer objects
tend to have a courser texture that do far away
objects (primarily due to perspective). Thus,
texture can be used as a cue to depth. Also,
further away subjects tend to be hazier than
close objects. As illustrated in Figure 7.31,
we can therefore use haze to infer distance. The
horizon also provides a cue to depth as we know
it is far away. Thus, objects closer to the line
of the horizon are perceived as being further
away (Figure 7.33).
43
Monocular Cues - Motion Parallax
As we move in our environment, objects closer to
us appear to move more relative to their
background than do objects far from us. For
example, as I move around the front of the class,
the position of the students close to me
relative to some point at the back of the class
moves much more that does the position of the
student at the back of the class. Thus, the
amount of motion parallax an object produces
can be used to judge its distance.
44
Depth Perception - Summary
So, we use a number of sources to infer depth
that fall under two general headings Binocular
Cues Convergence Retinal Disparity Monocular
Cues Interposition Perspective Texture Haze S
hading Proximity to Horizon Motion Parallax
Steve, address the question concerning the
nature/nurture of depth perception
45
Constancies of Visual Perception
As we move around our visual world is constantly
changing. Objects cast different images as we
move around them, lighting conditions change, the
retinal size of objects change as we move towards
and away from them, etc. However, we do not
notice all this. Instead, we form a
fairly stable perception wherein we do not
suddenly see everything as completely different
when a cloud goes in front of the sun. Our
ability to see a stable percept is due to certain
constancies in visual perception, the two most
prominent being brightness constancy and form
constancy.
46
Brightness Constancy
A piece of white paper looks white to us whether
we see it in sunlight or shadow.
We do not view the brightness of some area in
absolute terms, but rather we view it relative to
the brightness of the context Thus, when a cloud
passes over, everything gets less bright and, as
a result, nothing really seems to.
47
Form Constancy
When we approach and move around an object, we do
not see it getting larger and changing shape
although the retinal image is indeed getting
larger and changing shape. This seems to be a
somewhat top-down effect. We know how large
and the typical shape certain objects have. When
the size is smaller we assume the object is far
from us. When the shape is different, we assume
we are not viewing the object straight on.
48
Perception of Motion
We quickly and easily detect motion in our
environments. The text book describes some of
the studies of phenomenon relevant to motion
perception. I am not going to discuss these in
class. Rather, I leave this end part of the
chapter for you. Hopefully, it will move you.