Temporal Processing

Question 1

What is the temporal processing limit for human percept?

Answer 1

-Limits for human percept: 1-3 ms

Question 2

Why is measuring temporal resolution a difficult problem?

Answer 2

-Duration/BW tradeoff

• Solutions:
  1) Gaps in broadband noise
  2) Time-reversed broadband signals
  3) Temporal modulation transfer functions (TMTFs)

Question 3

How do measure TMTFs?

Answer 3

• Measure if the listener can detect which complex stimulus has a “modulation” in the amplitude
• Measure the “m” (modulation index, depth) at which a person can detect the modulation (20*log[m])

Question 4

How do you plot TMTFs?

Answer 4

• Y axis: threshold of the amount of modulation required for detection (20*log[m])
• X axis: modulation frequency (Hz)

Question 5

What do TMTFs show?

Answer 5

• Sensitivity to modulation is better at lower modulation frequencies
• Most important modulations: speech
• Most prominent modulation frequency in speech spectrum: 8 Hz

Question 6

What does gap detection in noise depend on?

Answer 6

• BW

- CF

Question 7

Describe gap detection in broadband vs. narrowband noise.

Answer 7

• Broader noise BW with an auditory filter with high CF
• Gap detection improves in broader noise BW because they result in faster fluctuations once passed through an auditory filter
• Faster fluctuations with lower dips make the gap in noise easier to detect than when there are slower fluctuations with higher dips
• Broader BW also results in less ringing which makes the gap easier to detect
• Therefore, broader BW and higher CF results in better temporal resolution

Question 8

Describe Eddins et al. (1992).

Answer 8

• Measured gap thresholds for noise bands as a function of BW
• Fixed upper cutoff frequency (UCF) but varied BW
• Gap thresholds improved with increased BW but were basically independent of UCF

Question 9

Why is the single filter hypothesis wrong?

Answer 9

• In Eddins et al. (1992), thresholds did not improve with increased upper cutoff frequency (UCF)
• Used across-frequency comparison (why thresholds improved with increased BW)

Question 10

Describe gap detection for tones of different phases.

Answer 10

• Standard-phase condition: same non-monotonicities as in Shailer & Moore (1987); when gap duration is an integer multiple of the signal period, the standard-phase condition is identical to the preserved-phase condition
• Reverse-phase condition: mirror image of standard-phase condition (still has non-monotonicities)
• Preserved-phase condition: performance monotonically improved with increased gap duration

Question 11

How did Shailer & Moore (1987) explain the non-monotonicities in gap detection for standard phase?

Answer 11

• Ringing in the auditory filter
• When the sinusoid is turned off at the start of the gap, the filter continues to respond
• When the gap duration is the period of the sinusoid, the sinusoid following the gap has the same phase as the ringing response
• With 2 extreme gap durations (short, long), the sinusoid following the gap is out of phase with the ringing response

Question 12

How does hearing loss affect gap detection for tones?

Answer 12

• TMTFs have same shape but occur at higher CBWs and at higher thresholds than for NH
• HL would be expected to have better performance (broader filters, less ringing, better performance than NH)
• Less non-monotonicities than with NH subjects
• Performance improves faster

Question 13

Describe expected gap detection in CI users.

Answer 13

• CI patients have total hearing loss, so any deficits in temporal mechanisms caused by cochlear damage should be maximized
• If HL causes retrocochlear deterioration of temporal resolution, we would expect CI patients (like HI) to have poor gap thresholds
• However, if gap detection were limited by cochlear processing, CI patients might have smaller gap thresholds than NH listeners because their nerves are stimulated directly with no mechanical resonance to limit the process
• Might expect better gap detection than NH because no ringing/smearing

Question 14

What are the effects of level on gap detection for CI users?

Answer 14

• Gap detection improves with increased level until a certain level at which threshold saturates
• Although there is variability, trend is monotonic

Question 15

Describe the sliding temporal integration window.

Answer 15

• A temporal window is a weighing function applied to the instantaneous energy of the signal over the time period covered by the window
• The window slides in time, so its output as a function of time represents a running weighed average of the energy of the signal (temporal excitation pattern)

Question 16

How does the sliding temporal integration window affect the internal representation of signals?

Answer 16

• Moore’s model of temporal resolution assumes that the internal representation of stimulus is “smoothed” over time
• Rapid temporal changes are reduced due to low-pass filter characteristic of TMTFs

Question 17

Describe duration discrimination experiments.

Answer 17

• Measure threshold separation (ms) as a function of either:
  1) time between tones of the same duration
  2) different duration of tones separated by same inter-stimulus interval

Question 18

What are some problems with duration discrimination experiments?

Answer 18

• Duration is not the only cue available
• Spectral confounds (i.e. splatter) are also available, resulting in additional sidebands in the frequency spectra in shorter duration tones
• Listeners could utilize the increased energy from the sidebands to detect duration differences, in addition to duration cues

Question 19

How do you perform a modulation detection interference experiment?

Answer 19

• Masker modulations are held constant; target modulations are varied
• Measure thresholds for modulation detection with and without maskers modulations

Question 20

What are the implications of modulation detection interference experiments?

Answer 20

• Supports the theory of a modulation filter bank

- Suggests that certain filters have greater sensitivity to modulation frequencies

Question 21

Describe age-related temporal processing deficits.

Answer 21

• Aging and hearing loss affect temporal processing
• Problems with older articles: age confounds
• Gordon-Salant et al. (2006) found that older people required increased temporal cues for word contrast discrimination than did younger people (OHI did even worse)
• Older people also perform worse with spectrally-degraded stimuli which force them to attend to temporal cues

Question 22

What are the blocks in Moore’s temporal resolution model?

Answer 22

1) Stimulus presented
2) Bandpass filter: half-wave rectification along BM (auditory filters)
3) Nonlinear device: compressive nonlinearities at the BM
4) Temporal integrator: “smooths” the signal based on weighted average of the output of BM compressive nonlinearities over a certain time window
5) Decision device: determined by task demands and output of temporal integrator