I. Definition of
psychoacoustics
A. Another term used for psychoacoustics is auditory
psychophysics
B. Psychoacoustics attempts to specify the relationships
between the physical characteristics of the sounds that
enter the ear (acoustics – 1st third of the class)
and the
sensations that they
produce (psychology)
II. Background of
psychophysics – famous people
A. Gustav Fechner (1801-1887)
1. Interested in studying the soul and felt that by
studying sensation,
he was studying the soul
2. Developed the basics
of psychophysical methods
used today
B. Weber – more later
C. S.S. Stevens – more later
III. Basics of psychophysics
A. Two approaches
1. Thresholds = measuring the limits of sensitivity
a. absolute
sensitivity
1) Ideal definition = smallest stimulus that can
be perceived
2) Realistic definition = the stimulus that
produces an
arbitrary level of performance
b. differential
sensitivity
1) Lowest difference in
stimuli which can be detected
2) Also referred to as just-noticeable difference
(jnd)
or difference limen (DL)
2. Scaling = ordering and distributing stimuli along a
perceptual dimension
IV. Classical threshold
measurement methods
A. Method of limits
1. Procedure
a. Descending
1) Start with a level that you think the listener
will
hear
2) If he hears that one, then present a lower level
3) Continue until the listener says he cannot hear
the
stimulus
4) Repeat
b. Ascending
1) Start with a level that you think the listener
will not
hear
2) If he does not respond, then present a higher
level
3) Continue until the listener says he hears the
stimulus
4) Repeat
c. Threshold taken as the average of the Y-N (or +/-)
transition
points between each series – Fig 7.1
2. Problems with method
a. Response – biases? à See homework question
b. Limited by step size used
3. Psychometric
function is how we plot results
a. Plot of percentage of yes responses as a function of
level
b. Figure 7.2
c. Threshold is an arbitrarily-defined percent correct
level
1) The book references 50%
2) You will also see many references to thresholds
in the 70 – 79% range
3) Remember there is a gradual improvement in
detection
with increase in level
4. Can be done to measure absolute or differential
sensitivity
B. Method of Adjustment
1. Procedure
a. Descending
1) Give listener control of the stimulus level
2) Instruct listener to adjust level until he cannot
hear
stimulus
3) Repeat
b. Ascending
1) Give listener control of stimulus level
2) Instruct listener to adjust level until he can
just hear
stimulus
3) Repeat
c. Figure 7.3
2. Problems
a. More response biases (homework)
b. Produces unreliable results
3. Can be done to measure absolute or differential
sensitivity
C. Method of Constant
Stimuli
1. Procedures
a. Choose stimuli of several values
b. Present these stimuli to listener in random order and
ask subject to respond when he hears it (or hears a
difference)
c. Record the proportion of times he reports hearing each
one at each level
d. Overhead (not in book)
2. Problems – homework
3. Of the classical methods, it is usually the method of
choice
4. Again, can be used for absolute and differential
sensitivity
III. Modern threshold
measurement methods
A. Theory of Signal Detection (TSD)
1. Purpose
a. Provides a rational basis from statistical decision
theory for
conceptualizing how subjects make
decisions under
conditions of stimulus
uncertainty
b. Allows separation of sensitivity to stimulus
differences or
ability to discriminate from various
kinds of
response bias due to prior knowledge,
stimulus
probabilities, payoffs, or other
motivational
factors
c. Differs from classical methods in that it assumes that
there is no such
thing as a "threshold" but that there is
a continuum which relates stimulus intensity to the
probability of
detecting that signal
d. Applicable to a wide variety of detection,
discrimination,
and recognition problems in many
areas of
psychology
2. Yes-no signal detection paradigm
a. Procedure
1) Some trials are catch trials à present no sound
2) Other trials are true trials where there is a signal
3) Ask person to indicate when they hear the signal
b. Possible outcomes - Fig. 8.1, Fig 8.4
1) Hit -
signal present/subject says yes
2) Correct
rejection - signal absent/subject says no
3) Miss -
signal present/subject says no
4) False alarm
- signal absent/subject says yes
c. Calculation
p(hit)
- p (false alarm)
p (hit)corrected = ---------------------------
1 - p(false alarm)
Corrects the hit
rate by factoring in chance
B. Adaptive methods
1. The level of the stimulus presented on a given trial
depends on the
subject's performance on previous trials
2. The subject must make a choice about which interval
contains a signal
a. 2AFC (2IFC) = 2 alternative (interval) forced choice
1) Subject is presented with two intervals one of
which
contains the signal
2) Subject pushes a button to indicate which interval
contained
the signal
3) The program looks at the response and decides
which level
to present next
a) A common set-up is to require two correct
answers
in a row before lowering the level and
one
incorrect answer to increase the level à
gives
you threshold at 70.7%
b) Another common set-up is to require three
correct
answers in a row before lowering the
level
and one incorrect answer to increase the
level à gives you threshold at 79.4%
a. 3AFC (3IFC) = 3 alternative (interval) forced choice
1) Similar to 2AFC except subject is presented with
three
intervals one of which contains the signal
V. Scaling techniques – a
brief introduction
A. Magnitude estimation
1. Developed by S.S. Stevens as a method for getting
people to scale the magnitude of stimuli such as the
loudness of sounds
2. Procedure
a. Tell subject to listen to sounds and have them assign
a
number to
reflect how loud it is
b. Present many signals and graph the number given as a
function of
signal level
c. Can be unrestricted – person is free to choose any
number that they
want
d. Can be restricted – tell person the range of numbers
that they may use (e.g. 1 to 100)
e. May or may not present a standard (modulus) sound
for subject to use as a baseline
B. Magnitude production
1. Subject controls level of sound and experimenter
controls numbers
2. Experimenter tells the subject to adjust the sound until
its
loudness is a certain number (e.g. 10)
C. Cross-modality matching
1. Proposed by Stevens but not really used until Hellman
and Meiselman in 1970s and 1980s (Hellman still uses
this technique)
2. Uses two sensory modalities (e.g. vision and audition)
3. Example: Tell subject to adjust line until it is as long
as
sound is loud
4. Very robust and can even be done by children as young
as four