Music 220c - MMN study on Musical Motives

Overview

My project, entitled "Encoding of staggered polyphonic musical motives: the effects of time-offset between motif entries on mismatch negativity (MMN)" is looking at a particular kind of brain response, observed in EEG, called the mismatch negativity (MMN). I adopted and made some changes to following Springer Reference definition for the MMN: The mismatch negativity (MMN) is an event-related potential (ERP, or brain response which is the direct result of a specific event) evoked in response to a perceived change in sensory stimuli. It is commonly elicited in an experimental paradigm in which a 'standard stimulus' is paired with a 'deviant stimulus', the standard being presented the majority of instances. The difference between the standard and the deviant stimulus may be in frequency, intensity, complexity, etc. The MMN generally occurs 110-150ms after the presentation of the deviant stimulus. The MMN is typically evaluated by subtracting the average brain response to the standards from the average brain response to the deviant.

Timeline

Progress Reports

Below I will be updating my progress on a weekly basis (newest first), as well as describing some of the challenges/trials/tribulations of the work

May 15th

All of the participants have been run, and the poster has been made in preparation for the conference I am presenting it at in a couple of weeks.

Below are the difference waveforms for all four conditions. What we are looking for here is a dip around 110-150 ms - this is the MMN. Conditions that have a greater dip have a stronger MMN. MMN also usually occurs in the front or center of the brain. Here we are looking at the front electrodes.
DiffWavAllCond You can see that the zero-overlap, one-voice condition tends to have the strongest MMN. This is what we expected. The fcl electrodes (front center left) has apparent MMN in the two-overlap two-voice condition. And the two-overlap one-voice condition shows weak MMN in the fc electrodes, which is also what we expected.

Generally, our conclusions from this work are that

May 6th

I have one more person today in the lab - Romain! - and then I'm done. Ge will be filming the capping of Romain so that I have documentation to present for my final project. I'm meeting with Takako on Friday to go over the data, so my processing of the data must be done by then. That's all for now!

April 29th

I had two people come into the lab so far, and are having another two today. The current number of participants I have scheduled before May 5th is 8, and by May 6th, I'll have 9! This is more participants than I could have hoped for given the time frame, and puts me more or less right on track with my schedule.

For next week, I'll have some documentation of the process in the lab, with a focus being on fitting and testing the cap.

April 18th

The stimuli are completed. Since this is a study of the processing of motives, the 'standard' stimulus is a brief, 5-note motive with the following contour:

5 note stimulus

The 'deviant' stimulus begins the same as the standard - the difference is that the 5th note ascends instead of desends.

There are four conditions overall:

  1. No overlap, one voice. In this condition, there is an eighth rest between motivic entries (like the above), and all stimuli are presented in the treble clef range
  2. No overlap, two voice. In this condition, there is an eighth rest between motivic entries, and presentation of the stimuli alternate between treble and bass clef range
  3. Two-note overlap, one voice. Here, a stimuli is only 3 notes into the 5-note sequence when another motive enters, and all stimuli are presented in the treble clef range
  4. Two-note overlap, one voice. Here, a stimuli is only 3 notes into the 5-note sequence when another motive enters, and presentation of the stimuli alternate between treble and bass clef range

Our hypothesis is that the MMN will be greater for conditions 1/2 than it is for 3/4, because the listener will have time to clearly tend to the end of the deviant, when compared to the overlap conditions, in which, by the time the 5th note of the deviant stimulus is occurring (which is actually the only different note of the deviant stimulus), the listener will already be hearing and following the next standard, so the MMN will be blunted! Two of our conditions are in two voices so we can test who people are understanding the two voices - are they tracing them separately? If so, there should be a strong MMN when looking at the brain waves in response to the lower voice and upper voices separately, perhaps with a stronger response in the upper voice (since typically, it is easier to hear/follow). If they are following them using the same 'memory trace', we would truly expect equal MMNs between the two voices.

Overall, there were certainly trials and tribulations when designing the stimuli. The following conditions had to be met/thought about:

Technically, the process for designing the stimuli included making maps of what stimuli and deviants can harmonically follow one another in the overlapping condition, writing out/troubleshooting the ordered stimuli in notation (I used Finale), creating .wav files of the stimuli using MATLAB scripts and .wav files of individual piano tones, and formatting another script to play the stimuli in the proper order in the Gentask software (the stimulus presentation software used in the Neuromusic lab). This entire process took me since the start of the quarter.