Does the introduction of an intermediate tone prevent/delay stream segregation between two tones?
by Cesar Duran
On Al Bregman's Auditory Scene Analysis page, it highlights Leo van Noorden's analysis of galloping patterns and their relation with streaming. In van Noorden's paper, participants were faced with a rhythmic 'galloping' pattern (High, Low, High, Silence). The first trial had a large difference in frequency, with the High tone being 1400 Hz and the Low tone being 500 Hz. As the cycle increased in speed, there was a segregation of streams, resulting in loss of rhythmic information. The second trial had a smaller difference in frequency, with the High tone being 1400 Hz and the Low tone being 1320 Hz. As the cycle increased in speed, the segregation of streams had occurred much later. The results of this experiment shed light on the mechanisms in the brain that group sound energy across time, also known as streaming. The results highlight the dependence of streaming on both time and frequency.
For my variation of this illusion, I chose to look at whether the gap between two frequencies can be bridged through the introduction of a third intermediate tone. I was curious to not only see if the addition of an intermediate tone would allow for a longer single stream, but also whether the optimal intermediate tone would be directly halfway between the two frequencies. My hypothesis was that the introduction of the third intermediate tone would provide a bridge that would make it more difficult for the streams to segregate, and the optimal frequency for the intermediate tone would be lower than the average of the two tones because the cochlea utilizes logarithmic scaling. I used 1500 Hz and 900 Hz for my two tones and had 5 trials where I varied the intermediate tone from {1000 Hz, 1100 Hz, 1200 Hz, 1300 Hz, 1400 Hz}.
Original Experiment:
1400 Hz Bridge:
1300 Hz Bridge:
1200 Hz Bridge:
1100 Hz Bridge:
1000 Hz Bridge:
Al Bregman, Auditory Scene Analysis. "Loss of rhythmic information as a result of stream segregation." 2008.
Comments
Anonymous
c) Thank you all for your comments! I had originally had the galloping pattern in place (ABCBA-Silence-ABCBA) but this led to less stream segregation due to the bridge in between, and more due to the fact that the pattern wasn't rhymic (The C's were every 6 beats, but the B and A had 2/4 beat alternating gaps). Therefore I adjusted the pattern to be more rhythmic, and as a result it lost some of the galloping characteristics that were seen in the original example. The speeding up done during the experiment was to show that stream segregation only occurs at higher rhythmic speeds.
d) In the future, I might adjust the question to account for the fact that it is no longer a galloping pattern, but rather a constant increased/decrease pattern.
0
Matthew Bradford
a) This is very interesting because it replaces the gallop with the rotations of a polygonal wheel. I mean that instead of hearing two opposing tones, you hear a sequence of different tones. The lower the bridge tone, the larger the sequence is.
b) I found that my experience during the demo did help in answering the question that was asked. The stream segregation did become worse as the bridge tone got lower. I would have preferred that all of the tones play at the same speed. It seemed like there was a confounding effect due to different ramps in speed between trials.
0
Michael Anoke
This was a very interesting experiment!
a. Listening to the different trials, I seemed to always segregate the lowest tone from the other ones until the last trial (1000Hz), and it was difficult for me to hear all three tones as a single stream. Except for the last trial, there seemed to be an "anti" gallop pattern that went BAB instead of ABA since I grouped the middle tone and the higher tone, and the middle tone occurred more frequently.
b. I believe that the experiment answered the question. It seemed like the intermediate tone made it easier to segregate the two tones (upper and lower tones) and greatly shortened how long I could hear the two tones as a "gallop". The "anti" gallop caused by the intermediate tone made the original ABA tone almost impossible to perceive.
0
Ben Radovitzky
Cool experiment! I noticed that the most "different' tone was always the one that would segregate first, but I also found it quite interesting how it was usually very easy to switch between hearing the noises as grouped to hearing them as segregated, which was a bit bizarre. The only ones which I noticed were very difficult to keep grouped as opposed to segregated were the very first and last examples, with the 1400Hz and 1000Hz bridges. However, in each case, all of the "galloping" parts were never segregated within each other.
I would say that this experiment highlighted an answer to a slightly different question than was posed (if I'm understanding it correctly). As I stated before, the galloping parts never really seemed to lose their grouping, however the added intermediate tone was very easy to segregate. I feel like I am left with the question of whether there is something about the relative distance in frequencies between these tones that make the segregation more salient, and why it is that adding a more "different" intermediate tone makes it much harder to segregate within the galloping noise. Very interesting and thought provoking!
b) Now, pretend you are the researcher and assess whether your experience of the demo was helpful in answering the stated question.
Make sure to include your name so we can grade you. Feel free to include comments on other illusions as well.
