Re: Monaural vs diotic stimulus presentation (Erick Gallun )


Subject: Re: Monaural vs diotic stimulus presentation
From:    Erick Gallun  <gallun(at)SOCRATES.BERKELEY.EDU>
Date:    Sun, 6 May 2001 01:40:31 -0700

>When designing a psychophysical study of a monaural phenomenon, that >is, one in which binaural information does not play a part, I've >noticed that almost everyone presents their (headphone) stimuli >monaurally. I was wondering if anyone can give me good reasons for >not presenting the stimuli diotically, that is, the same stimulus >presented to both headphone channels. >Specifically, I'm designing a study on temporal gap detection, using >noise markers of various bandwidths. I'm interested in relative >performance as a function of the noise marker composition. So, if the >absolute gap detection thresholds across the board were lower with >diotic than monaural presentation, due to some sort of summation >effect, it wouldn't matter to me. I think you will find that your subjects do better with diotic than monaural stimuli. I presented data at ARO in february showing better performance with diotic than monaural presentation in an intensity increment detection task (as have others with quite different tasks - such as signal detection in noise). My current thought on this (and I'd be happy to get feedback on it) is that signals with no external noise are improved more by diotic versus monotic presentation since the noise at the two ears is mostly the result of variability in neural firing, which is uncorrelated between the two ears. Neural variability has a small effect when there is binaurally correlated external noise in the stimulus, since the variability in the signal due to the noise is dominated by the external noise - in other words, the signals at the two ears are highly correlated and thus the second produces very little additional information. Neural variability is most apparent when it is the only noise source present - for example, when the signal is a pure tone.. In that case, it results in essentially independent samples of the signal. In fact, my data suggested a "square root of two" improvement, which is the prediction from independent noise sources. Since you are not studying a diotic phenomenon, it shouldn't matter if performance is elevated in all conditions. And, as you suggest, subjects prefer diotic to monaural presentation. I would be interested to hear if there is a better reason - aside from saving money on earphones. erick gallun grad. student hafter lab, UC Berkeley


This message came from the mail archive
http://www.auditory.org/postings/2001/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University