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

       

1