This is a talk presented in an interdisciplinary workshop named "The Melodic Mind" by Dr. Daniela Sammler at Max-Planck Institute for Human Cognitive and Brain Sciences.

1. Intracortical myelination in
musicians with absolute pitch
Seung-Goo (“SG”) Kim & Thomas R. Knösche
Max-Planck Institute for Human Cognitive and Brain Sciences
“The Melodic Mind” organized by Dr. Daniela Sammler
2015-08-24

2. Overview
• Introduction: absolute pitch & myeloarchitecture
• Methods: image acquisition, processing, & inference
• Results: effect of absolute pitch
• Discussion: anterior superior temporal gyrus and ventral
pathway

3. Overview
• Introduction: absolute pitch & myeloarchitecture
• Methods: image acquisition, processing, & inference
• Results: effect of absolute pitch
• Discussion: anterior superior temporal gyrus and ventral
pathway

4. What is absolute pitch (AP)?
• Ability to recognize the name (or pitch chroma) of any given tone (“C4” or
“G#3”) without external reference
• Rarely acquired ability (7% of Western musicians, 30% of East Asian
musicians [1])
• Related to early musical experience (4~7 year-old [2]), fixed-Do training [3]
• Uncontrollable and instant (e.g., Stroop effect; “Sol” as “Do” in G major
[4,5])
• Very accurate at pitch chroma but some of them are not so good at pitch
height (e.g., octave errors) [1]
[1] Miyazaki, 2004 [2] Zatorre, 2003 [3] Willson et al., 2012 [4] Itoh et al., 2005 [5] Schultz et al., 2013

5. • Early perceptual categorization
: smaller right-PT [1], larger right-
HG [2]; activation in left-PT during
passive listening [3,4]
• Late cognitive association to
verbal/non-verbal representation
: thicker IFG [6]; activation in right
DLPFC [2-5]
• Frontotemporal connectivity
: phase synchrony [7]
DLPFC
[1] Keenan et al., 2001 [2] Wengenroth et al., 2014 [3] Ohnishi et al., 2001 [4] Willson et al., 2009
[5] Zatorre et al., 1998 [6] Dohn et al., 2015 [7] Elmer et al., 2015
Two components: macroscopic findings
HG
PT
http://clipartbest.com/

6. But how does the AP work?
• What does distinctive functional activation patterns imply about
the neural mechanism of the AP processing?
• How does macroscopic morphology (e.g. GMD, CT, ROI volume,
and so on) contribute to the AP recognition?
• Microscopic investigation powered by ultra-high-field MRI is
expected to helpful for better understanding of the AP
mechanism.

7. Intracortical myelination
• Long-range connections
• fasten axonal transmissions
• decrease variability of transmission
speed thus increasing synchrony
• Still on local connections
• reduce cross-talk so that increasing
specificity
• prevent neuroplasticity after critical
period [1]
Cyto- Myelo-
Vogt.1903
[1] McGee et al., 2005

8. In-vivo myelin mapping
Wallace et al., 2002 Dick et al., 2012
ML
S
I

9. Overview
• Introduction: absolute pitch & myeloarchitecture
• Methods: image acquisition, processing, & inference
• Results: effect of absolute pitch
• Discussion: anterior superior temporal gyrus and ventral
pathway

10. Participants
• 8 AP musicians (5 women) & 9 non-AP musicians (5 women)
• Age (20~40 y.o) and handedness (≥ 70) matched (p < 0.39)
• Ethnicity and training onset age were not matched (5 Asian
musicians with AP, the others were Europeans; Asians started 3
years earlier)
• Self-claimed AP possession was screened by a web-based AP
test and later reconfirmed by pitch identification test

11. Behavioral tests
• Pitch identification test using muted digital piano (72 pure tones
+ 72 piano tones across 3 octaves) spacing at least [1]
• Frequency discrimination threshold (FDT) estimated by a
staircase procedure (1-up/2-down) with about 100 trials [2]
• Melody part of the Musical ear test (MET) to match musical
aptitude between AP and non-AP musicians [3]
• FDT and MET were matched between groups (p<0.11)
[1] Miyazaki 1989 [2] Micheyl et al., 2006 [3] Wallentin et al., 2010

12. Image acquisition
• MP2RAGE (7T), 0.7-mm-isovoxel; Head-coil system: RAPID 8-ch.
TR / TE / TI1 / TI2= 5000 / 2.45 / 900 / 2750 msec
FA1/FA2= 5/3 degrees; dimension: 320 x 320 x 240 voxels
1st inversion 2nd inversion Quantitative T1 Uniform contrast

13. Image processing
• Cortical surface reconstruction using FreeSurfer
• qR1(=1/qT1) values sampled directly from 0.7-mm volume at 25,
50, 75 % of cortical thickness along outward normal vectors

14. qR1(1=1/qT1) values over layers

15. Statistical inference
• SurfStat MATLAB toolbox used for 2-D random field theory
(RFT)-based family-wise error rate (FWER) control at cluster-
level (alpha=0.05)
• Iterative surface-based smoothing corresponding to a 2-D
Gaussian kernel with a FHWM of 7 mm
• Inferior regions (such as temporal pole, fusiform gyrus)
excluded from search region for signal dropout in 7T images

16. Overview
• Introduction: absolute pitch & myeloarchitecture
• Methods: image acquisition, processing, & inference
• Results: effect of absolute pitch
• Discussion: anterior superior temporal gyrus and ventral
pathway

17. AP-test confusion matrix

18. qR1 examples

19. qR1
= β0
+ β1
age+ β2
sex + β3
ethnicity + β4
AP +ε

21. qR1
= β0
+ β1
age+ β2
sex + β3
AP +ε

22. Inwardprojectionby25,-50,-75%
ofcorticalthickness
qR1
= β0
+ β1
age+ β2
sex + β3
ethnicity + β4
AP +ε

23. qR1
= β0
+ β1
age+ β2
sex + β3
ethnicity + β4
AP + β5
(−logFDT)+ε

24. Overview
• Introduction: absolute pitch & myeloarchitecture
• Methods: image acquisition, processing, & inference
• Results: effect of absolute pitch
• Discussion: anterior superior temporal gyrus and ventral
pathway

25. • Early perceptual categorization
: smaller right-PT [1], larger right-
HG [2]; activation in left-PT during
passive listening [3,4]
• Late cognitive association to
verbal/non-verbal representation
: thicker IFG [6]; activation in right
DLPFC [2-5]
• Frontotemporal connectivity
: phase synchrony [7]
DLPFC
HG
[1] Keenan et al., 2001 [2] Wengenroth et al., 2014 [3] Ohnishi et al., 2001 [4] Willson et al., 2009
[5] Zatorre et al., 1998 [6] Dohn et al., 2015 [7] Elmer et al., 2015
Two components via dorsal pathway
PT
http://clipartbest.com/

26. • PT is sensitive to:
spatial information, speech
perception, acoustic pattern
analysis, music perception [1]
• PP is sensitive to:
identity of auditory object [2],
voice identity [3], physical
property of vocal tract [4],
salience of pitch [5],
and pitch chroma [6]
PP
PT
Planum temporale vs. planum polare
HG
[1] Griffiths & Warren, 2002 [2] Zatorre et al., 2004 [3] Belin et al., 2001 [4] Kriegstein et al., 2006
[5] Penagos et al., 2004 [6] Warren et al., 2003
http://clipartbest.com/

27. Pitch chroma & pitch height
•
Scientific pitch notation: A4 = 440 Hz
• Pitch chroma: 12 tones consist 1 octave
e.g. A, A#, B, C, C#, …, G#
• Pitch height: integer index of octave
e.g. (0, 1, 2, …, 8) for 88-key (full scale) piano
Warren et al., 2003https://en.wikipedia.org/wiki/Piano

28. Pitch chroma vs. pitch height
Warren et al., 2003

29. Octave errors
• APs make errors in pitch height with almost no errors in pitch
chroma) suggesting distinctive processes of pitch chroma and
pitch height [1-4]
[1] Miyazaki, 1988 [2] Deutsch, 2013 [3] Deutsch & Henrhotn, 2004 [4] Takeuchi & Hulse, 1993

30. • APs’ behaviors shows higher
sensitivity to pitch chroma [1]
• PP is sensitive to pitch chroma [2]
and related to auditory object
recognition [3,4]
• And greater cortical myelin in the
right PP in APs [5], which prohibits
plasticity after the critical period [6]
• Conjecture: “Recognition of pitch
chroma may occur as an auditory
object recognition”
DLPFC
PP
PT
Ventral pathway for pitch chroma
HG
[1] Miyazaki, 1998 [2] Warren et al., 2003 [3] Zatorre et al., 2004 [4] Kriegstein et al., 2006 [5] Kim et al., 2015
http://clipartbest.com/

31. On-going analyses for further evidence
• Functional/anatomical
connectivity
Resting-state functional connectivity
seeding from the right PP
(our data)
Non-AP (n=1) AP (n=1)
Correlation[-1,1]
Octave-tuned voxels in non-musicians
(Moerel et al., 2015)
• Tonotopy mapping to find pitch
chroma sensitive voxels

32. Thank you for attention!