Re: Invitation to criticize dissertation idea. (Al Bregman )

Subject: Re: Invitation to criticize dissertation idea.
From:    Al Bregman  <bregman(at)HEBB.PSYCH.MCGILL.CA>
Date:    Tue, 16 May 2000 16:19:54 -0400

Dear David and list. In response to your query to the list, I think that it's important to study which aspects of timbre most strongly affect perceptual grouping, but I wouldn't want to put it as a competition between pitch and timbral variables. There are two reasons: 1. Whereas pitch has an underlying physical variable (repetition rate of the waveform) that has a reasonably well defined "magnitude", and each physical variable underlying timbre is also translatable into a magnitude, you don't know how to compare the two magnitudes, a priori. So if you want to argue, for example, that some physical variable, T, which causes timbral differences, is more powerful than another one, P, which causes pitch differences, you have to know, a priori, what size difference in T is equal to a given difference in P, PERCEPTUALLY. Otherwise a critic might simply say that you varied T over a wider perceptual range than you varied P. In other words, he would argue that the range of T differences in this experiment was larger than the range of P differences, and that is why T seemed to overcome P in controlling segregation. 2. Some of the timbral variables (such as noise bandwidth and mistuning of harmonics) actually affect perceived "pitch STRENGTH" (the degree to which the listener experiences a definite pitch) even after you adjust the center frequency for any change in pitch VALUE; so the pitch strength is not the same for different values of timbre. For example, pitch strength weakens as you increase the bandwidth of narrow-band noises, holding center frequency constant. Therefore you cannot really vary pitch properties and timbral properties independently, as you would have to do if you wanted to see which controlled grouping better. This problem arises because of the choice of technique, in which the pitch and timbre of every tone must be controlled so as to create a set that independently varies both of them. A design that varied only one variable at a time (e.g., a typical streaming experiment) wouldn't run into this problem. From your mention of onset asynchrony, I as [[portion lost - dpwe]] overlapping or synchronous tones. I would want to include a wholly synchronous condition as well -- one in which the ascending sequence was interleaved in the spaces between the descending one. Wasn't this the format of the original scale illusion reported by Diana Deutsch? Best of luck, Al Bregman ----- Original Message ----- From: David Spondike <Dspondike(at)AOL.COM> To: <AUDITORY(at)LISTS.MCGILL.CA> Sent: Monday, May 15, 2000 7:35 PM Subject: Invitation to criticize dissertation idea. > Dear List: > I am new to your cadre, so let me introduce myself and pose my > question/request. I am a doctoral candidate and Kent State, working on ... > responses as to where you think research is needed, and what might > be a good topic for a doctoral dissertation. Please keep in mind that I am a > musician, not a cognitive psychologist (though I did my minor in psych). > Thanks to all of you who took the time to read this, and of course, to > all of you who wish to respond. > > David Spondike > Dspondike(at)

This message came from the mail archive
maintained by:
DAn Ellis <>
Electrical Engineering Dept., Columbia University