A Multimodal Study of InKey vs' DistantKey Music Sequences

1
A Multimodal Study of In-Key vs. Distant-Key
Music Sequences
MEG Group Vijay Venkatraman, University of
Pittsburgh, Bioengineering Tim Ambrose University
of Pittsburgh, Medical School Vanja Kljajevic,
Ph.D. University of Groningen
Faculty Dr. Bill Eddy TA Erika JCL Taylor

• July 31, 2008

2
Why study processing of music sequences?

• Recent research has shown overlap in syntactic
  processing in language and music, e.g.
• Positive-going brain potential P600 (Patel et
  al., 1998)
• MEG evidence on the involvement of Brocas area
  (Maess et al., 2001)
• These findings argue against the strict
  modularity hypothesis, indicating involvement of
  shared resources in syntactic processing in music
  and language (Patel, 2003 2007)

3
Hypotheses

• 1. Processing of music syntax relies on the
  resources that support syntactic processing in
  language.
• 2. Syntactically more complex structures require
  more processing resources.  
•  

4
Research Questions

• Are syntactically more complex sequences such as
  Distant-Key harder to process than In-Key
  sequences
• Do they require more time for integration
  processes?
• Do they elicit more activation in the areas
  implicated in processing?

5
Experimental Design

• Compare processing of In-Key with Distant-Key
  sequences

6
Participants and Procedure

• Simultaneous MEG/EEG acquisition
  n4 Elekta Neuromag
• 1kHz sampling rate, no online filtering
• Structural MRI acquisition
  n3 Siemens 3T Trio
• MPRAGE, TR2.2 secs, TE 3.43ms,
  192256176,pixel resolution 1 mm1mm 1mm

NOTE - We also completed two language studies,
results from which are not presented here
7
Behavioral Results
Not significant
8
Acquisition Steps
Also measure distance of head from helmet (HPI)
9
Analysis Procedure
10
Coregistration Sphere Fitting
1
11
Coregistration Sphere Fitting
1
Spatial Filtering
2
RAW
SSS
12
Coregistration Sphere Fitting
Frequency Filtering
1
3
Bandpass 0.5 - 40 Hz
Spatial Filtering
2
RAW
SSS
13
Analysis Procedure (cont.)

• 4) Eliminate artifacts (ocular, muscle, etc.)
• 5) Average data by condition
• 6) Re-reference and Grand Average ERP data
• 7) Transform head position of participants MEG
  data into common distribution

Accepted Trials Per Condition (total possible
60 per)
Vs.
14
Source Localization Strategy
15
(No Transcript)
16
Slight right hemispheric dominance?
Left
Right
Grand Average of Distant-Key gt In-Key (4
Subjects) - Distributed Source
17
(No Transcript)
18
How do we evaluate quality?
Goodness of fit of variance that this dipole
explains in this ROI 95 Confidence Volume we
are 95 sure the dipole belongs in a mm3
area Comparison to Forward Solution (see below)
BLUE original RED predicted by forward
solution
19
Focus 3 timepoints/regions
20
t 91ms
GROUP Distant-Key
INDIVIDUAL Distant-Key
ERP No clear effect
MEG
R
21
t 252 ms
GROUP Distant-Key
INDIVIDUAL Distant-Key
ERP Frontal Effect
------ Distant Key ------ In Key
MEG
R
22
t 600 ms
GROUP Distant-Key
INDIVIDUAL Distant-Key
ERP (Central Parietal) P600
------ Distant Key ------ In Key
MEG
R
23
fMRI Experiment
24
Learning more from fMRI

• Why do fMRI?
• What regions will fMRI activate? WE DID THIS
• Use the fMRI results to drive MEG source
  localization.
• Could we compare time course in fMRI and MEG for
  regions of interest.

Paradigm An event related design with 30 Trials
per condition. fMRI acquisition
n3 subjects
Siemens 3T Trio EPI TR2secs, TE30ms, 646435,
pixel resolution 3.125 mm3.125 mm3.1
Randomized Jitter 2,4,6 secs
Music
Response / Response Feedback
25
fMRI Analysis
Analysis The analysis was using standard SPM 5.
Motion Correction
Coregister (MR to Func)
Normalization (using gray matter)
Smoothing (FWMH-8mm)
Group Analysis
3 subjects
Individual Subjects (HRF)
Used motion parameters as covariates.
26
fMRI Results

• (Distant Key gt In Key) showed Brocas Area
  homolog (BA44,BA45) activation with plt0.01 and
  k40 voxels.
• Other activate regions were the BA 3, BA40
  (parietal regions).
• Replicates previous studys activation of Brocas
  area for syntactic in music. (Maess et al., 2001)

27
fMRI Results
We find the left Insula for Distant-Key condition
with p lt0.01 and k40 voxels. Our MEG results
showed right insula for the Distant-Key condition
at t252 ms.
We find bilateral temporal area activation for
distant Key with plt0.01 and k40 voxels similar
to our MEG results (right temporal) at t91ms for
distant key.
Do we have activation differences in MEG and fMRI
results? YES
28
Why the difference?

• The fMRI experiment was run using mono headphones
  and MEG was run using stereo headphones. We
  think the lateralization differences are due to
  this.
• The difference could also be due to the signal
  source for the BOLD and MEG signal ?

29
Discussion

• Differences in processing of In-Key and
  Distant-Key conditions found with regard to when
  and where in the brain syntactic integration
  processes occur
• The findings are in tune with the recent research
  on musical and language syntax in indicating
  involvement of Brocas area and P600 in syntactic
  integration processes (Patel et al., 1998, Maess
  at al., 2001 Grewe et al., 2006 Santi
  Grodzinsky, 2007)
• Additional findings of the current study (e.g.
  the role of insula in syntactic integration
  processes) are also congruent with language
  syntax processing evidence (Loevenbruck et al.,
  2005)
• Thus, the current data provide further support to
  the hypothesis that syntax in music and language
  share resources (Patel, 2003)

30
Acknowledgements

• Dr. Ani Patel, for providing music stimuli
• Dr. Seong-Gi Kim, for very helpful comments and
  encouragement
• Dr.Anto Bagic and Anna Haridis (MEG Center)
• Elisabeth Ploran, for help with fMRI experimental
  design
• Jeff Phillips
• Denise Davis and Dr.Costin Tanase (MR Center)
• Fellow 2008 MNTP trainees
• Tomika Cohen
• Rebecca Clark
• Sorrentos Pizza ?

Special thanks to Dr. Bill Eddy Erika JCL Taylor
31
Additional Slides
32
(No Transcript)