Perception

The acquisition and processing of sensory

information so that we can see, hear, taste,
touch and smell the world.
 Visual Perception
Making sense of the visual world seems
easy, but seeing is actually very complicated

Indeterminacies: A given pattern of light on

the retina might be caused by a number of
actual objects out there in the world, so how
do we know which to perceive?
Necker Cube
Shape Indeterminacy
Size Indeterminacy
 Brightness Indeterminacy

The amount of light coming off a piece of

paper if you move it from the sunlight to
a shadow changes dramatically

Yet we perceive it as the same color…

Because the visual perceptual system

makes automatic
assumptions/adjustments
6
 Illusion caused by
automatic adjustment for
brightness indeterminacy

7
 Illusion caused by
automatic adjustment for
brightness indeterminacy

8
9
 Some Assumptions to Deal with
Brightness Indeterminacies
• Surfaces are uniformly colored

• Gradual changes are often caused by

shadow

• Light comes from above

Assumption to deal with
Shape Indeterminacy?
 Gestalt Law of Pragnanz

We tend to see the simplest possible

organization.
 Gestalt Laws of Perceptual
Organization, examples
(a) the law of proximity
(b) the law of similarity
(c) the law of good continuation
(d) the law of closure.
Displays showing conflict between…

(a) proximity and similarity (50/50)

(b) shape and color
(c) shape and color (color wins)

All adapted from Quinlan and Wilton (1998).

 Size Indeterminacy - Solutions
• Cues in the Environment – These are often referred
to as monocular cues because they don’t require
both eyes
• Familiar Size:
• For example, if a child appears larger than an adult, it is likely
that the child is closer to the observer.

• Occlusion: An object that occludes another is closer

• Texture Gradient: A field is assumed to have a uniform

texture gradient, so if more detail is visible in part of the
field, it is assumed to be closer
15
 Size Indeterminacy - Solutions
o Relative Height: Objects higher in the picture plane are farther away

o Atmospheric Perspective: Objects in the distance look less distinct with

a bluish tinge as they are viewed through dust & water particles that
scatter light

16
 Size Indeterminacy - Solutions
o Linear Perspective: Parallel lines converge in the
distance, so the closer they are to converging, the
farther away the location

17
 Size Indeterminacy - Solutions
So visual system makes assumptions to deal with these
indeterminacies
Visual illusions take advantage of these assumptions…

Ponzo Illusion
Which assumption does this
exploit?

18
 What is this?

A
How do you know?
 Pattern Recognition
• How are we able to so easily recognize certain
patterns?

aAaAaA Aa

• We must match visual

information with
 Pattern Recognition
• Template Theories

We have stored versions (templates) of every pattern we

can recognize (a) and when we see a pattern out there, we
compare it to all our templates and come up with closest
match

But the enormous variation in visual stimuli that all match

one common template is difficult for these theories
 Pattern Recognition
Feature Theories
We match features of visual patterns to sets of features
stored in memory, rather than specific templates
“two lines slanting together to a point connected by a crossbar” =
A

Now, the angle, orientation, size or exact shape of

different As (capital anyway) won’t be a problem.

In your head, determine if each letter string has a “Z” in it

(yes/no) and raise hand when done with list.
Ready…
How would template theories explain the difference?

Support for feature theories of pattern detection

 Pattern Recognition
Feature Theories

Suggest that we focus on specific, smaller features first,

and then more global processing occurs

First we look for a point, and a crossbar, and then we

recognize the whole A

Let’s look at some support for feature theories from the brain…
Visual input
Hubel & Wiesel, 1959
orientation columns ocular dominance
columns
From primary visual cortex of monkey. These are
trained/tuned by experience, not there at birth,
kittens raised in dark for a few months, blind for life.
An anaesthetized
monkey, injected with
radioactively labeled
glucose, viewed a
flickering bulls-eye
pattern. The glucose
was taken up by active
neurons. The animal was
then sacrificed, and V1
was surgically removed
and flattened. The
flattened V1 was then
used to expose
radioactively sensitive
film. V1 maintains a
retinotopic map.
 Pattern Recognition
• Feature Theories
According to pure feature theories, a cell should respond
to the feature it “likes” (bar/moving bar in specific
place/direction) the same regardless of what is around it.

But, Weisstein & Harris (1974) showed that these cells

responded more strongly when the liked features were part
of a coherent form.
Navon (1977)

2 task types: congruent

1.ID big letter
2.ID small letter

2 trial types:
3.Congruent
4.Incongruent
incongruent

Big letter incongruency interfered with little letter ID more than other
way around
Perception not always features 1st, then whole later (specific to
global) – context matters!
Watch this!
RODENT
What was in
the 3 spot?
rd

D or E?
SVKAFD
What was in
the 3 spot?
rd

A or K?
 Word Superiority Effect
(we perceive words more easily than non-word letter strings)

Top Down Influence

Why does this matter?
 What is this?

How do you know?

If template theory is at least partly right, how do we deal with recognition
from different perspectives? A card for every possible perspective!??!
 Recognition-By-Components
Biederman (1987)
• We have a mental toolbox of about 36
different basic shapes (blocks, cylinders,
cones, spheres, arcs wedges…) called
geons

• Everything we
see can be
broken down
into component
geons
Viewpoint Dependence/Independence
Biederman claimed that geons were
viewpoint independent:
can be perceived/recognized equally
easily from any perspective

This helps answer the “how do we

recognize bicycles from all different angles
when they make such different shapes?”
Viewpoint Dependence/Independence
• Object recognition seems to be viewpoint
independent for more simple recognition
tasks
• Is it a car or a chair?

• But viewpoint dependent for harder tasks

• Which siblings face is it?
Viewpoint Dependence/Independence
• Evidence for both comes from cognitive
neuroscience
• Inferotemporal cortex (IT),
one place where visual
information goes after occipital
cortex, and where object
recognition takes place

• Some neurons in IT fire only

when you look at bike form
specific viewpoint
• Viewpoint dependent

• Others fire when you look at

bike from any perspective
• Viewpoint independent
 Object Recognition Disorders
Cognitive neuropsychology
AGNOSIA – INABILITY TO RECOGNIZE OBJECTS

• Apperceptive AGNOSIA
• Early visual perceptual processing deficits
• Patients cannot draw copies of things they are looking at

• Associative AGNOSIA
• Perceptual processing in tact
• Deficits in accessing relevant knowledge about objects in
long term memory
• Can copy objects – but still cannot recognize them
 Object Recognition Disorders

Agnosia patients do not fall neatly into one or the other of

these categories (apperceptive or associative).

• Some cannot even tell the difference between a

triangle and a rectangle
• Some can recognize objects when they are alone
(carrot on a blank table), but not mixed in with other
objects (in a salad)
• Some show that they do not use top down information
• Look for a piece of fruit in this picture
• Eye tracking shows most of us look first in the fruit bowl, on the table, etc.
• Particular agnosia patient did not look in the obvious spots
 Face/Object Recognition Disorders

Agnosia: https://www.youtube.com/watch?v=ze8VVtBgK7A
Prosopagnosia: https://www.youtube.com/watch?v=vwCrxomPbtY
Face Recognition
A holistic process
 Prosopagnosia:
Inability to recognize faces
• Usually caused by brain damage to specific area in
temporal lobe – fusiform face area – not JUST faces

• Some prosopagnosics
can recognize objects

• Other patients can

recognize faces but not
objects

• Suggests…
 Visual Imagery
Helps us to plan actions by visualizing the results
 Visual Imagery
Imagery resembles perception
Kosslyn - Visual imagery is like “seeing with the mind’s eye”
– Mechanisms of imagery are the same as mechanisms of
perception
– Visual images are arranged in space the way pictures or drawing
are – not propositional, e.g., hat ON TOP of head

• Charles Bonnet Syndrome patients actually confuse

visual imagery and perception
– They feel as though they are actually perceiving things that have
no corresponding stimulus in the real world
 Visual Imagery
Imagery resembles perception

• Imagery can FACILITATE visual perception

– Binocular rivalry study
• Different images to each eye – percept switches back and forth normally
• Which is perceived first is influenced by an initial common (to both eyes)
stimulus AND initial visual image about equally

• Imagery can INTERFERE with visual perception

– Ratings of vividness of visual or auditory images was
affected most by visual or auditory secondary tasks,
respectively
 Visual Imagery
Imagery resembles perception – BRAIN IMAGING
• Same part of the brain lights up in visualization
and perception of stuff (color, faces, etc.)
– Early visual perceptual processing areas in occipital
cortex are involved in imagery, more so for imagining
fine detail
– Vertical and horizontal edge detectors are engaged in
imagining vertical and horizontal edges respectively
– Over all, about 2/3 overlap between brain areas used for
perception and imagery

• Knocking out early visual processing areas with

rTMS (repetitive transcranial magnetic
stimulation) impairs imagery task performance
 Visual Imagery
Imagery does not resemble perception
• Images must be totally generated by top down
processes, in the absence of any bottom up,
perceptual input
• Imagine a cube balanced on one corner and cut
across equator – what is shape of crosscut?
– Images consist of simplified versions of reality

• Double dissociation in brain damaged patients

• Which means…?
 Visual Imagery
Imagery does not resemble perception
Double dissociation in brain damaged patients

• Damage to L temporal lobe

– patients can perceive images
just fine, as evidenced by
copying ability

– But cannot imagine images –

unable to draw these simple
images from memory

– L temporal lobe is important

for knowledge of concepts,
including objects
 Visual Imagery
Imagery does not resemble perception
• Double dissociation in brain damaged patients

• Anton’s Syndrome
– Blind patients who have
almost all of visual
processing cortex destroyed,
but have visual images so
strong they mistake them for
percepts
 Visual Imagery
Imagery does not resemble perception
• Rotate image 90 degrees clockwise in your mind,
what do you see in each?
 Visual Imagery
Imagery does not resemble perception

• Images cannot be used as

flexibly as percepts
 Mental Rotation
BUT, we can MANIPULATE visual images, and we do…
 Mental Rotation

Are the 2 shapes the same?

• A?

• B?

• C?
 Mental Rotation
• What does this
tell us about
rotating mental
images?
 Mental Rotation
• Mental rotation is
an analogue
process…

• Rotating an
image in your
mind is just like
rotating an object
through actual
space
 Mental Rotation

Gender effects

• One of the few

places where gender
effects are pretty
solid, BUT…

• Difference can go away if instructions suggest women are better

at task than men – socialization?
• Boys play more video games, which helps MRT ability
• Women are better at object location
– Evolutionary perspective – men hunt and need to get home, women gather and need to
find stuff
 Mental Rotation

• Involves brain areas used for

– SPATIAL processing - parietal cortex
– MOTOR planning and activity - posterior frontal cortex

• Notice these are different from the occipital regions involved in

static imagery
 Inattentional Blindness
Focusing attention on the white team, 50% don’t notice gorilla.
But only 27% miss it if they are watching the black team.

There is much that we “see” but

do not “perceive”
Unconscious (subliminal) perception
Subjective threshold
• People report being unaware of a visual stimulus

Objective threshold
• People’s performance on tests shows no evidence of
having perceived a stimulus

Naccache et al. (2002)

– People judged whether a number was below or above 5
– Just before number, another congruent or incongruent number
was flashed for 29ms
– People never noticed (below subjective) and couldn’t guess
above chance whether flashed number was above/below 5
– BUT they were faster for main task on congruent trials
Unconscious (subliminal) perception
Rees (2007)
– Presented houses or faces below all thresholds
– By looking at brain activity, researchers could guess what image
had been shown with 90% accuracy

– Non-consciously perceived things are processed to considerable

extent by the visual system.

object
face
 Blindsight

Patients with damage to

visual processing areas –
occipital cortex - have no
conscious access to visual
information presented in
the damaged visual field

But if asked to “guess” or

make motor response
(point to dot, or hold hand
at orientation of line), the
information is in there