EFFECTS OF DURATION OF FRINGE OF NOISE AND FRINGE OF SILENCE ON DETECTION OF BRIEF SIGNALS MASKED BY NOISE.
Item
-
Title
-
EFFECTS OF DURATION OF FRINGE OF NOISE AND FRINGE OF SILENCE ON DETECTION OF BRIEF SIGNALS MASKED BY NOISE.
-
Identifier
-
AAI8103965
-
identifier
-
8103965
-
Creator
-
TZUO, PI-FONG LIN.
-
Contributor
-
Eli Osman
-
Date
-
1980
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Psychology, Experimental
-
Abstract
-
This dissertation is an attempt to further our understanding of the temporal processes involved in the masking of brief signals by noise. The masker was broadband white noise of constant spectrum level (about 46 dB SPL), and its temporal character in relation to the signal was the main independent variable. The signal was either a 400-Hz tone-burst or a broadband noise-burst. Its duration was fixed at 11.5 msec with 2.5 msec rise/fall times. The masking noise was always presented throughout the 11.5 msec signal interval and was otherwise manipulated as follows. In the 'noise-fringe' conditions the masker was symmetrically extended forwards and backwards in time with respect to the signal interval. In the 'gap-fringe' conditions, silent gaps were similarly introduced symmetrically in time on the two sides of the signal interval, in the otherwise continuously presented masker. Special attention is commanded by the two extreme conditions, usually referred to as 'gated' and 'continuous' masking. Nine one-sided fringe durations ranging from 0 to 95 msec were chosen for the masker settings for each type of fringe and each of the two signals. Psychometric functions were obtained from each of three listeners using a single interval Yes-No procedure. Each psychometric function was converted to a plot of an index of detection (d') in log units vs. signal-to-noise ratio in dB. Regression analysis then yielded threshold (at d' = 1) and slope parameters for each such function. Analysis of the joint pattern of changes in threshold and slope is the basis for interpretation of the study.;A brief summary of the results follows. Continuous masking is more effective than gated masking for the tonal signal, but not for the noise signal. For both signals, as the noise-fringe is increased from the gated condition, the threshold first rises along a negatively accelerated course to peak in the vicinity of 25 to 45 msec. Subsequently it drops gradually out to 95 msec, and drops again for the continuous masking condition. As the gap-fringe is introduced the threshold first remains stable, then drops to a minimum at 15 msec, and shows smaller changes thereafter. Slope functions of noise-fringe duration and gap-fringe duration are also similar for both signals. For noise-fringe durations of 0 to 15 msec, the slope remains stable somewhere between .60 to .85, after which it begins to rise gradually, soon crosses unity, and keeps increasing to a maximum less than 2. As the gap-fringe is introduced and expanded, the slope drops smoothly, crosses unity between 10 to 15 msec, then drops to around .90, and finally decreases to the level of the gated condition.;A general statistical decision theoretic model of a leaky power integrator was derived and applied for the special case of an exponential weighting function. The model provides a reasonable explanation of the data for short noise-fringe and long gap-fringe durations, but generally must be supplemented by other detection processes. Thus, at very brief noise-fringe durations for tones, results may be due to phase-delays. For relatively long noise-fringe durations, results may be consequences of increasing temporal uncertainty (which soon saturates) and decreasing masking of large signal peaks, both of which are consequences of masker transients. Selected parameters of the model and particular features of the results were related to the concept of a 'temporal critical masking interval'.;One major contribution of this research is the demonstration that important restrictions on theoretical interpretations of psychophysical results are implied by analysis of complete psychometric functions.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
-
Program
-
Psychology
