Scribd
Upload
32 views

Lectures Perception

The document discusses graphical illusions and how the brain processes visual information. It provides examples of different types of illusions, including geometric illusions, perspective illusions, size illusions, color illusions, and more. It also discusses principles of visual perception like Weber's law and Stevens' law, and how these laws can help design more effective graphs that are easier to interpret.

Uploaded by

vikram
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
32 views

Lectures Perception

The document discusses graphical illusions and how the brain processes visual information. It provides examples of different types of illusions, including geometric illusions, perspective illusions, size illusions, color illusions, and more. It also discusses principles of visual perception like Weber's law and Stevens' law, and how these laws can help design more effective graphs that are easier to interpret.

Uploaded by

vikram
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 66

Graphical Perception

Visual Processing And The Brain


• When an image is passed to the brain it is processed
through increasingly complex steps until it reaches the
higher areas of the brain.

• Very little is known about the detail of these processing


steps, but some intriguing information comes from
visual illusions.

• Illusions appear to to be a result of the brain deriving


incorrect information about what is being viewed.

• In more interesting illusions, the brain derives


conflicting information about the scene being viewed.
The Cafe Wall Illusion
The Cafe Wall Illusion
• The lines in the image are either horizontal or vertical.

• The illusion stems from the fact that part of our visual
processing system is tells us that the lines must be
sloping while the cognitive part of our brains tells us
they are not.

• This illusion was first noticed by a perceptual


psychologist strolling past a group of workmen tiling
the wall of a cafe in Bristol, UK.
The (Real) Cafe Wall Illusion
Types of Illusion
• Illusions come in a variety of forms.

• Geometric illusions are among the most interesting.


Such illusions include:

– distortions of lengths
– distortions of angles
– distortions of areas
– distortions of shapes
Perspective Illusions
• A number illusions result from the visual system
interpreting visual elements as giving information about
depth.

• Perspective gives us very strong depth cues.

• Misinterpreting geometric features as resulting from


perspective produces strong distortions.
A Perspective Illusion
Another Perspective Illusion
The Ponzo Illusion
The Müller-Lyer Illusion
A Possible Source Of The Müller-Lyer Illusion
Angle Distortion Illusions
• The next group of illusions shown how angles can be
misperceived by our visual systems.

• This is of some interest in presentation graphics.


The Poggendorf Illusion

Which of the two lower lines


is the extension of the upper line?
The Poggendorf Illusion

Which of the two lower lines


is the extension of the upper line?
Explanation
• It appears that we have a tendency to overestimate the
size of acute angles and to underestimate the size of
obtuse ones.

• This effect is shown in the one of the Yarbus


experiments shown in the last lecture.

• A variety of illusions result from this misjudgement.


The Zöllner Illusion 1
The Zöllner Illusion 2
The Ehrenstein Illusion
Size Illusions
• Perspective cues can lead us to misjudge the size of
objects.

• This is not the only way in which we can be mislead


about the sizes.

• The are a number of size illusions which happen in


purely two-dimensional diagrams.
The Horizontal-Vertical Illusion
The Horizontal-Vertical Illusion
A Line Length Illusion
The Jastrow Illusion
Tichener Illusion
Colour Illusions
• There are many illusions related to colour.

• Many of these are due to the fact that we do not see


colour in isolation, but relative to its surrounding colour.

• Often this can be explained as a result of lateral


inhibition.
An Irradiation Illusion
Mach Banding
The Hermann Grid
Elke Lingelbach’s Variant of the Hermann Grid
Simultaneous Contrast I

The appearance of colours depends on their surroundings.


Simultaneous Contrast II
Simultaneous Contrast III
Graphical Perception
• Some visual processing takes place without any
concious effort on our part.

• Psychologists call this preattentive vision.

• In the context of extracting information from graphs we


will call it graphical perception.

• Graphs which convey their information at this


unconcious level allow us to extract the information
without any concious effort on our part.

• Such graphs are said to provide inter-occular traumatic


impact (this is a joke).
Graphical Cognition
• Some visual processing requires that we conciously
inspect the things that we are looking at.

• Psychologists refer to this kind of activity as cognitive.

• It is graphical cognition which allows us to make


statements like

– The largest person is third from the right.


– The steepest slope in the graph is near x = 4.5.
Graphical Design
• Graphs can affect us at either the perceptual or cognitive
level (or both).

• The most effective (and eye-catching) graphs always


operate to some extent at the perceptual level.
Graph Drawing
• When we draw a graph we encode a numerical value as
a graphical attribute.

• When we look at a graph the aim is to decode the


graphical attributes and extract information about the
numbers which were encoded.

23.7 24
31.3 Encode Decode 31
19.5 19
41.0 40
27.4 26

• If possible we should arrange for the decoding to


happen at the perceptual level.
Designing Effective Graphs
• To design effective graphs we must know which
graphical attributes are most easily decoded.

• We need a selection of possible graphical attributes and


an ordering of their ease of decoding.

• There are a number of possibilities.


Encoding Using Angles
Encoding Using Circle Areas
Encoding Using Irregular Areas
Encoding Using Lengths
Position on a Common Scale

● ●
● ● ●
Position on Identical but Unaligned Scales

●
●
●
●
Encoding Using Slopes

● ● ●

● ●
Encoding Using Luminance
Perception “Laws”
• Perceptual psychologists have established a number of
empirical laws for perception.

• The two most important laws hold very generally and


apply to many graphical encodings.
Weber’s Law I
Weber’s law applies to a variety of perceptual encodings, but
we will apply it to length.

• Consider two lines with lengths x and x + w.

• If w is very small, there is only a very small chance that


we will notice that the lines have different lengths.

• As w gets larger, the chance of detecting a difference


increases.

• Weber’s law says that the chance of detecting a


difference depends on the value of w/x.
Weber’s Law II
• For a given individual the difference between x and
x + w will be detected with probability px (w).

• Now turn this around. For a fixed p, let the value of w


which is detected with this probability be

wp (x)

• Weber’s law states that

wp (x) = kp x

• Since wp (x) ∝ x, we detect relative differences in


values.
Weber’s Law III
• We detect a 1cm change in a 1m length as easily as we
detect a 10m change in a 1km length.

• Weber’s law appears to hold for many different


graphical encodings.
Stevens’ Law
• Suppose that we encode a value x and then decode it to
obtain a perceived value p(x).

• Stevens’ law says

p(x) = Cxβ

• The values of C and β depend on both the encoding


method used and the observer.

• This law holds for many encodings.


Power Coefficient and Nonlinearity
• If the relationship p(x) and x is nonlinear, we can be
mislead in what we infer about the x values.

• The linearity of the relationship between p(x) and x


depends only on the value of β, with linearity if and
only if β = 1.

• Here are some typically observed ranges for β.


Length: 0.9 – 1.1
Area: 0.6 – 0.9
Volume: 0.5 – 0.8
Stevens’ Law – Perception of Large Values
• Consider area with β = 0.7.

• Suppose that we compare an area of size 2 with an area


of size 1. The perceived ratio of areas is:

p(2) 20.7
= 0.7 = 1.62
p(1) 1

• We don’t see the bigger area as twice as large.


Stevens’ Law – Perception of Small Values
• Consider area with β = 0.7.

• Suppose that we compare an area of size 1/2 with an


area of size 1. The perceived ratio of areas is:

p(1/2) 0.50.7
= 0.7 = 0.62
p(1) 1

• We don’t see the smaller area as half the size.


Perception of Area and Volume
• Our perception of areas and volumes is conservative.

• When values are presented as areas or volumes, we


underestimate the large values relative to the small ones
and overestimate the small ones relative to the large
ones.
Combining Weber’s and Stevens’ Laws
Consider encoding using length with β = 1 and area with
β = 0.7. When we compare the values x and x + d we
perceive the relative value
x+d d
=1+
x x
for length and the relative value
0.7
(x + d)0.7

d 0.7d
= 1+ ≈1+
x0.7 x x

for area.
We are more likely to detect small differences when a length
encoding is used than when area encoding is used.
Ranking Graphical Encodings
• Cleveland and McGill carried out an extensive study of
graphical encodings to obtain a best to worst ranking.

• The rankings they examined were

angle, area, colour hue, length,


colour brightness,
position (common scale),
position (identical unaligned scales),
colour purity,
slope, volume
One Cleveland and McGill Experiment
• 7 graphical encodings

• 3 judgements per encoding

• 10 replications per subject

• 127 experimental subjects

• Assessment criterion:

error = |judged% − true%|


One Cleveland and McGill Experiment
1 2

A B C D B

3 4

A A

B B

C C

D D

5 6

A A

B B

C C

D D

D
Results

Position (Common)

Position (Nonaligned)

Length

Angle

Slope

Circle Area

Blob Area

8 10 12 14
The Encoding Ranking
1. Position on a common scale

2. Position along identical, non-aligned scales

3. Length

4. Angle / Slope

5. Area

6. Volume

7. Colour properties
Recommendations
• Use the highest possible encoding on the
Cleveland-McGill scale.

• The preferred encodings are:

– Position on a common scale.


– Position on identical, unaligned scales.
– Length.

• Be careful when using angles and slopes when encoding


numerical values.

• Don’t use area or volume to encode numerical values.

• Don’t use colour to encode numerical values.


Perception In Maps
Our perception of the size of areas is affected by the shape of
those areas. Here Georgia is bigger than Florida, but doesn’t
look it.

GA

FL
Angular Encoding in a Pie Chart
Pie charts have a very weak perceptual basis.

D
Encoding Using Length
Encoding Using Identical Unaligned Scales

You might also like