Chapter 3 & 4 of Moore (1997)

現在位置: ROM首頁 / 聽覺心理學 / Chapter 3 & 4

Chapter 3 & 4 of Moore (1997)
Frequency Selectivity (III) & Temporal Processing

Scope

Models to Account for CMR

Cross-filter comparison: Detecting the disparity in the modulation pattern across different filters.
Dip-listening: The modulation envelope gives away the best times to listen for the signal.
Subjects can discriminate coherent and incoherent modulation of a pair of noise bands (Richards, 1987).->Supporting the first model.
Hall & Grose (1987): Maskers=on-freq. + flanking bands. Signal= on-freq. (No across filter disparity, only a hange in overall level)->Sizable CMR->Neither factor is necessary.
Grose & Hall (1989): Masker=series of tones. Signal=one of the tone. Signal onset: min or max in the modulation.->Got CMR only at min.-> For dip-listening.
Sometimes the detection of a modulation on one carrier freq. would be interfered by a modulation on another CF. Chap 4.

Profile Analysis

One can do across-filter comparison for stimuli without envelope fluctuations (Green, 1988).
Stimuli: Series of tones. Task: Detect an increment in the level of the central component. Overall level of the hole stimulus varies over a wide range.
See p. 127 for the most effective conditions for PA.
Moore thinks that PA can be considered as the temporal version of CMR. I'm not sure I'll take his arguments here.
But I agree with him that PA is old wine in a new bottle. We'll came back later when talking about timbre.

Forward & Backward Masking

FM, how does it differ from adaptation and fatigue? Masker duration (short) and 200ms time window after mask offset.
BM is experience-dependent, FM is not.
Properties of FM:(1) Delay-dependent (Fig. 3.17, left). (2) Rate of recovery is greater for higher masker levels. (3) See Fig. 3.17, right for the growth function of the masker level. (4) FM increases with the duration of masker till 50-200ms. (5) Masker & signal must be related in frequency.

Possible Sites for the Occurrence of Forward Masking

Ringing of the BM.-> Low freq. should be greater than high freq. Why?
Short-term adaptation or fatigue of auditory nerve or higher centers in the auditory system. Not plausible in view of the physiological data.
Persistence of neural activity evoked by the masker. Will be back in Chap 4.

Evidence for Placing the Site at BM

The growth function of the masker level can be explained by the input-response function of the BM (See Fig. 3.18 & right panel of Fig. 3.17).
The shallow part of the growth function is due to signal level falling within the low level steep rising phase.
The rising part of the growth function is due to signal level falling within the mid level flat phase.
Predictions: When mask freq.=signal freq., and the delay is short, the growth function takes a slope of 1. Look at Fig. 1.11. When MF < SF, signal is compressed but masker is not.-> Steep growth function should be found.

Lateral Suppression Revealed by FM

Simultaneous masking will not reveal LS, because LS affects both signal & masker (Houtgast, 1972).
Presumptions: (1) Suppression occurs no later than FM. (2) Masker-induced suppression must have decayed by the time signal is delivered.
Repeated-gap masking technique:

Lateral Suppression Revealed by FM (continued)

When the masker is a high-pass or a low-pass noise and the signal is well within the pass band, the masking effect is weaker than when the signal is at the notch edge. (Not so in SM)
n It is argues that there is less lateral suppression around the frequency edges (in the neural representation of the masker).
Continuity effect & pulsation threshold. PT is believed to be able to map out the excitation pattern of the masker, including the effects of linear filtering & LS.

If you need a print friendly verison of this document, you can download the .DOC file.