Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements
description
Transcript of Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements
![Page 1: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/1.jpg)
Revised estimates of human cochlear tuning from
otoacoustic and behavioral measurements
Christopher A. Shera, John J. Guinan, Jr., and Andrew J.
Oxenham
![Page 2: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/2.jpg)
Background
• Key characteristic of hearing: frequency tuning of cochlear filters– Sensory cells respond to a preferred range of
energy– Filter bandwidth 1/ sharpness of tuning
![Page 3: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/3.jpg)
Background
Assessments of cochlear tuning
• Non-human mammals– ANF recordings in live anesthetized animals
• Humans– Psychophysical measures
• Masking procedures
• Pure tone detection in background noise
![Page 4: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/4.jpg)
Downfalls
• Assumptions underlying pure tone detection method are uncertain
• Physcophysical detection tasks depend on filter characteristics as well as neural processing
• No way to validate behavioral measures in humans
•Humans
–Psychophysical measures
•Masking procedures
•Pure tone detection in background noise
Authors believe that human cochlear tuning has been underestimated
![Page 5: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/5.jpg)
Aims
• Compare current measures of human cochlear tuning with animal measures
• Develop a noninvasive measure of cochlear tuning based on otoacoustic emissions
• Test correspondence between physiological and behavioral measures of frequency selectivity
![Page 6: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/6.jpg)
Aims
• Compare current measures of human cochlear tuning with animal measures
• Develop a noninvasive measure of cochlear tuning based on otoacoustic emissions
• Test correspondence between physiological and behavioral measures of frequency selectivity
![Page 7: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/7.jpg)
Determination of bandwidth
QERB
• Measure of “sharpness” of tuning based on critical bandwidth
• QERB(CF) = CF/ERB(CF)
Smaller bandwidth = higher QERB
Frequency
Le
vel (
dB
SP
L)
SignalMasker
Auditory filter
2 kHz
![Page 8: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/8.jpg)
Results
Genuine species differences or erroneous human data?
![Page 9: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/9.jpg)
Aims
• Compare current measures of human cochlear tuning with animal measures
• Develop a noninvasive measure of cochlear tuning based on otoacoustic emissions
• Test correspondence between physiological and behavioral measures of frequency selectivity
![Page 10: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/10.jpg)
Experiment II• Subjects
– Guinea pigs (n=9)– Cats (n=7)– Humans (n=9)
• Measure stimulus-frequency otoacoustic emissions (SFOAEs)
– Cochlear traveling waves scattered by the mechanical properties of the cochlea– Recordable sounds emitted from the ear– Evoked by a pure tone
• Calculate SFOAE group delays (NSFOAE)– Negative of slope of emission-phase vs frequency
![Page 11: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/11.jpg)
Theory
• NSFOAE = 2(NBM)Normalized emitted wave delay is double the normalized BM
transfer function delay
• NBM= delay of BM transfer function• NSFOAE = emission group delay
Can use measurable NSFOAE group delays to estimate NBM
![Page 12: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/12.jpg)
Traveling wave delays
![Page 13: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/13.jpg)
Theory II
• At low levels, smaller bandwidths (larger QERB) correspond to steeper phase slopes (longer delays)
• BM tuning at low levels nearly identical to ANF tuning so:
QERB NBM ==> QERB = kNBM
Where k is a measure of filter shape
![Page 14: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/14.jpg)
Application
• Use measurable SFOAE emissions to estimate NBM
• Use NBM to estimate QERB using known k values from other species
![Page 15: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/15.jpg)
Results
![Page 16: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/16.jpg)
![Page 17: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/17.jpg)
If this is right, it suggests: 1) Human k is a factor of
3 larger than in animals
2) Human QERB is very different from cats and guinea pigs
![Page 18: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/18.jpg)
If this is right, it suggests: 1) Previous measures
underestimate human filter “sharpness”
2) Such sharp tuning may facilitate speech communication
![Page 19: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/19.jpg)
Aims
• Compare current measures of human cochlear tuning with animal measures
• Develop a noninvasive measure of cochlear tuning based on otoacoustic emissions
• Test correspondence between physiological and behavioral measures of frequency selectivity
![Page 20: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/20.jpg)
Experiment III• 8 Normal-hearing humans
• Detection of a sinusoidal signal– 10dB above threshold in quiet– Frequencies: 1,2,4,6,8 kHz– 5ms after offset of burst of masker
• Frequencies: 2 .25f wide spectral bands of Gaussian noise placed 0, 0.1, 0.2, 0.3, 0.4 f below signal frequency
– gated by 5ms raised-cosine ramps
• Measured thresholds using 3-alternative forced-choice procedure
• Use mean data to derive cochlear filter magnitude responses
![Page 21: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/21.jpg)
Reasoning behind methodology
• Use low, near threshold tuning curves – Avoid compression & non-linear affects
• Noise masker extends spectrally above and below signal frequency– avoid off-frequency listening – avoid confusion between masker & signal
• Non-simultaneous masking– Minimize suppressive interactions between masker and
signal
• Constant signal level (instead of masker level)– paradigm used in neural threshold measurements
![Page 22: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/22.jpg)
Results
![Page 23: Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements](https://reader036.fdocuments.us/reader036/viewer/2022070419/56815ad9550346895dc8a5e1/html5/thumbnails/23.jpg)
Conclusions• Human cochlear filters are substantially sharper than
commonly believed• Contrary to prior beliefs
– Human Q filters are not constant above 500Hz– Human tuning may be sharper than cat – Human and cat tuning may vary similarly with CF
• Supports the assumption that k is invariant across species
• Suggests revised understanding of the cochlear frequency-position map