Psych 480: Fundamentals of Perception and Sensation

1
Psych 480 Fundamentals of Perception and
Sensation

• Dr. Keith S. Jones

2
Why do we perceive?
3
Realism
Idealism
vs.
The Ecological approach
Information Processing approaches
Gibson
Elaboration?
Yes
No
Constructivist approaches
Computational approaches
Helmholtz
Rock
Marr
4
A general IP model
Distal Stimulus
(a lot can go wrong here)
Proximal Stimulus
(if elaborative)
Transduction
Processing
Percept
Cognition
(if elaborative)
5
Psychophysics
Distal Stimulus
How do these relate to one another?
Proximal Stimulus
(if elaborative)
Transduction
Processing
Percept
Cognition
(if elaborative)
6
Activity (2 pts)

• How would you measure the magnitude of a
  perception?
• Describe your measurement technique.
• Explain how it accomplishes your goal.
• What problems, if any, exist with your technique?

7
Early attempts

• Gustav Fechner (1860)
• Concepts
• Absolute threshold
• Difference threshold
• aka JND
• Webers Law
• ?I/IK
• Methods
• Method of constant stimuli
• Method of limits
• Method of adjustment

http//www.uni-leipzig.de/fechnerday/
8
Fechners attempt

• His measurement system
• Use a persons absolute and difference thresholds
  as a counting system.
• Assumes that
• All JNDs are subjectively equal
• Webers law is true
• Fechners law
• Perceived magnitude K log Intensity

9
Example

• Fechners law
• Perceived magnitude K log Intensity
• Assume
• absolute threshold 50 units
• difference threshold 10.

10
Problems

• Webers law is NOT always true
• It breaks down at the extremes
• Stevens (1957) argued that JNDs are not
  subjectively equal.
• He argued that they might be for metathetic
  dimensions (e.g., pitch, color), but not
  prothetic dimensions (e.g., loudness, brightness)

11
The tower of Babel was never finished because the
workers could not reach an understanding on how
they should build it my psychophysical edifice
will stand because the workers will never agree
on how to tear it down. (Fechner, 1877)
12
Stevens attempt
Exponent .5

• Methods
• Magnitude estimation
• Magnitude production
• Cross-modality matching
• Stevens Power Law
• Perceived magnitude K Intensity power

brightness
Exponent 1
line length
Exponent 3.5
electric shock
13
Problems

• If you use a category rating scale (i.e., a
  likert scale), then data follow a logarithmic
  function, not a power function.
• Stevens argues that these scales are inherently
  biased b/c the categories are not subjectively
  equal.
• There is also evidence to suggest that magnitude
  procedures are affected by context as well.
• Exponent inconsistency over time.

14
The debate rages on ...

• Many applied settings have adopted Stevens Power
  Law as a standard.
• However, from a theoretical standpoint, work is
  on-going.
• Krueger, L.E. (1989). Reconciling Fechner and
  Stevens Toward a unified psychophysical law.
  Behavioral and Brain Sciences, 12, 251-320.

15
New psychophysics
Signal Detection Theory
ß
If likelihood ratio (Psn/Pn) gt ß, then person
says Yes, else he or she says No
http//acad.cgu.edu/wise/sdt/sdt.html
16
Signal Detection Theory
17
Signal Detection Theory
Correct Rejection
False Alarm
Hit
Miss
18
Calculating d and ß

• Based on the p(hit) and p(false alarm), you can
  calculate d.
• d zp(fa) - zp(hit)
• Based on the height of the curves, you can
  calculate ß.
• ß height(hit)/ height(fa)

19
How it works ...
d zp(fa) - zp(hit)
ß
Zhit -.5
Zfa 2.5
20
How it works ...
ß height(hit)/ height(fa)
ß
Height .34
Height .05
21
Rules of Thumb

• For d, if it is
• d lt 1.5 difficult
• d 1.6 to 2.5 moderately difficult
• d 2.6 to 3.5 moderately easy
• d gt 3.5 easy

• For ß,
• three or less is common
• If greater than three, person is conservative

22
Tests of TSD

• Vary signal probability
• Should change ß w/o changing d
• Pay-off matrices
• Systematically affect ß by changing operators
  strategies w/o changing d
• Vary quality of signal
• Should affect d w/o affecting ß
• 2nd chance experiment
• If person is incorrect, then must be below
  threshold so giving a 2nd choice shouldnt
  matter. If TSD is correct, then 2nd choice
  should beat chance b/c people differentially
  weight different possibilities.