NEURAL CORRELATES OF VOCAL PITCH REGULATION IN SINGING

NEURAL CORRELATES OF VOCAL PITCH REGULATION IN SINGING JEAN MARY ZARATE Dept. of Neurology & Neurosurgery Mc. Gill University

INTRODUCTION ¡ Precise vocal pitch regulation necessary for speech and song ¡ Vocal pitch regulation requires integration between: l l l ¡ Stable vocal motor system Auditory feedback Interface between these two components not well-understood Used singing to find neural substrates for audio-vocal integration

Elicit learned vocalizations Initiate vocalizations

EXP 1: Experience-dependent neural substrates involved in vocal pitch regulation (Zarate & Zatorre, 2008) ¡ 12 non-musicians (6 ♀), 12 singers (6 ♀) ¡ HYPOTHESES: l SIMPLE: basic network for singing (Perry et al. , 1999) IGN: ↑ attention areas, ↓ auditory cortical activity COMP: audio-vocal integration = ACC, STG, insula? l Singers: l l Singing tasks: singers > non-musicians ¡ Experience-dependent modulation in basic network for singing, audio-vocal integration ¡

SIMPLE – PERC (SINGER ∩ NON-MUS) M 1 Cbl SMA Th STG/ INS PAC/ STG ACC 6. 1 y = -14 x=0 z=0 2. 5 IGN – SIMP (SINGER > NON-MUS) PT/ STG STS 4. 0 y = -22 2. 0 d. PMC 3. 7 z = 68 2. 0 SINGER > NON-MUS > SINGER COMP – SIMP (GROUP DIFF) p. STS RCZa 4. 4 x = -10 x = 50 2. 4

EXP 1: KEY FINDINGS ¡ IGN: non-mus had pitch-shift responses l l ¡ Pitch-shift response = vocal stabilization system Training needed to suppress stabilization Audio-vocal integration: ¡ Non-mus: d. PMC (sensorimotor association) ¡ Singers: RCZa, p. STS

EXPERIMENT 2: Neural networks involved in voluntary and involuntary vocal pitch regulation in experienced singers (Zarate et al. , submitted) ¡ ¡ 9 singers (6 ♀) SIMPLE; IGN/COMP 200 c and 25 c pitch shifts l l ¡ COMP 200 c = voluntary vocal pitch regulation Pitch-shift response in IGN 25 c = PAG? Unable to verify role of PAG due to temporal resolution limitations of f. MRI

FUNCTIONAL CONNECTIVITY: COMP 200 EFFECTIVE CONNECTIVITY: COMP 200 (vs. SIMPLE) p. STS seed EFFECTIVE CONNECTIVITY: IGN 200 (vs. SIMPLE) p. STS seed

¡ EXP 1 & 2: RCZa, p. STS, anterior insula l l l ¡ Recruited after vocal training Functionally connected to each other p. STS interacts with IPS to monitor feedback EXP 3: Training effects in non-musicians (Modulation of functional network for singing after auditory training) l l l Better auditory skills = better vocal accuracy? Better vocal accuracy modulations in singing networks Melodies: ¡ Singing tasks: 50 c & 100 c melodies, simple singing ¡ Perception: micromelody discrimination (<100 c interval)

FUNC. CONNECTIVITY (POST – PRE) right PT seed

EXP 3: CONCLUSIONS ¡ Short-term auditory training l l l ¡ training effects with micromelody discrimination no training effects on vocal production no neural modulations specifically induced by training-enhanced vocal production Dissociation between perceptual and production skills? l l different time-courses of behavioral improvement auditory-motor training necessary

¡ ¡ Consolidated after adequate audiovocal training Short-term auditory training does not engage or consolidate network

ACKNOWLEDGMENTS Robert J. Zatorre Advisory Committee: D. Louis Collins Alan Evans David Ostry Université de Montréal / BRAMS / CIRMMT: James Bergstra Douglas Eck Sean Wood Mc. Gill / MNI: Pierre Ahad Patrick Bermudez Marc Bouffard André Cormier Karine Delhommeau Michael Ferreira Nicholas Foster Talya Grumberg Funding: ¡ Canadian Institutes of Health Research (CIHR) ¡ Eileen Peters Mc. Gill Majors Fellowship ¡ Centre for Interdisciplinary Research in Music Media and Technology (CIRMMT) New York: Henry Mc. Donagh III Members of the Z-Lab

FUTURE DIRECTIONS ¡ A-V network specific to vocal pitch? l l manipulate other features (e. g. , formants) training effects: foreign language students MEG, EEG/ERP: pitch-shift response ¡ Auditory training vocal accuracy ¡ l l ¡ more testing sessions of vocal production longer auditory training Similar network with other perturbations? l somatosensory feedback

EXP 1: Audio-vocal integration SINGERS & NON-MUS EXP 2: Voluntary/involuntary vocal pitch regulation SINGERS EXP 3: Vocal pitch regulation after auditory training NON-MUS

¡ ¡ SIMPLE: Sing back single note PITCH-SHIFTED TASKS: ignore/compensate for ± 200 c-shift

IGNORE COMPENSATE

SIMPLE – PERC (SINGER ∩ NON-MUS) M 1 STG/ INS SMA Cbl Th PAC/ STG ACC 6. 1 x=0 2. 5 y = -14 z=0

IPS SMG 4. 2 y = -50 2. 0 SINGER > NON-MUS CONJUNCTION IGNORE - SIMPLE STS PT/ STG 4. 0 2. 0 y = -22

COMPENSATE – SIMPLE ACC IPS 3. 9 SMG 2. 0 y = -44 d. PMC 3. 7 z = 68 2. 0 SINGER > NON-MUS x=4 NON-MUS > SINGER CONJUNCTION r. ACC x = -10 p. STS 4. 4 2. 4 x = 50

EXP 1: Experience-dependent neural substrates involved in vocal pitch regulation (Zarate & Zatorre, 2008) ¡ Behavioral tasks: l l ¡ SIMPLE, IGN: singers > non-mus COMP: both groups successful Programmed to stabilize systems against disturbances Training needed to suppress stabilization mechanisms f. MRI results l l l SIMPLE: singers ≈ non-musicians COMP/IGN: ↑ auditory activity in singers Audio-vocal integration: ¡ Non-mus: d. PMC ¡ Singers: r. ACC, p. STS

IGNORE pitch-shift response Responses to long pitch shifts (>500 ms): – Early: ~100 -150 ms, more automatic – Late: ~300 ms, may be subject to voluntary control

EXPERIMENT 2: Neural networks involved in voluntary and involuntary vocal pitch regulation in experienced singers (Zarate et al. , submitted) ¡ ¡ 9 singers (6 ♀) SIMPLE; IGN/COMP 200 c and 25 c pitch shifts HYPOTHESES: l l Resp. magnitude: COMP 200 c > IGN 200 c Singers cannot suppress pitch-shift responses to small shifts: COMP 25 c = IGN 25 c IGN/COMP 200 c networks similar to exp 1 PAG pitch-shift response in IGN/COMP 25 c?

EXP 2: CONCLUSIONS ¡ Pitch-shift responses to IGN 25 c under less voluntary control than IGN 200 c l ¡ Role of PAG in pitch-shift response: not verified l l ¡ part of stabilization system occurs in milliseconds, f. MRI temporal resolution in seconds MEG, EEG/ERP: temporal interaction during A-V integration Voluntary vocal corrections: same network for different magnitudes: r. ACC, p. STS, anterior insula l l functionally connected to each other p. STS interacts with IPS to monitor shifted feedback

EXPERIMENT 3: Modulation of functional network for singing after auditory training (Zarate et al. , in prep) HYPOTHESES: l Auditory training with pure tones ¡ ↑ micromelody discrimination (pure- and vocal-tone) ¡ ↑ vocal accuracy l Melodic singing requires audio-vocal integration: ¡ similar regions seen in Exp 1, 2 ¡ auditory working memory (e. g. , inf. frontal) l Modulation of regions after training: ¡ singing network ¡ audio-vocal integration

• Perception: – 2 micromelodies: same/different? – Trained/tested with micromelodies (pure & vocal tones) • Production: simple singing & 5 -note melodies – Middle note ≈ 250 Hz – Intervals: 50 and 100 cents

EXP 3: ORDER OF TASKS beh pre Trained 9 subj (6 ♀) Control 10 subj (6 ♀) Production: ¡Simple ¡Melodies Perception: ¡Micromelod y discrimination f. MRI pre Production: ¡Simple ¡Melodies TRAINING f. MRI (2 weeks) post YES NO Production: ¡Simple ¡Melodies beh post Perception: ¡Micromelody discrimination Production: ¡Simple ¡Melodies

SIMPLE – PERC (PRE) SMA PAC / STG / PT INS sensorimotor (mouth) ACC 7. 8 2. 5 x=2 y = -16 z = 10 MEL(50+100) – SIMPLE (PRE) PT/STG 6. 6 2. 6 y = -12 z=4

POST – PRE

SCANNER PARAMETERS Exp 1 & 3: 1. 5 Tesla ¡ TR = 10 s, TE = 85 ms ¡ voxel = 5 mm 3 ¡ 25 slices (whole head) ¡ Matrix: 64 x 64 Exp 2: 3 Tesla, cardiac gating ¡ TR = 10. 3 s, TE = 60 ms ¡ voxel = 3. 5 mm 3 ¡ 40 slices (whole head) ¡ Matrix: 64 x 64

DUAL-STREAM MODEL OF AUDITORY PROCESSING ¡ Rauschecker/Tian 2000: l l Ventral: “what” – auditory object info Dorsal: “where” – auditory spatial info Belin/Zatorre 2000: Dorsal = “how” ¡ Warren et al. 2005: Dorsal = “do” ¡ ¡ Updated model: Dorsal = how / do

SINGING NETWORKS IN OTHER STUDIES

Hickok et al. 2003: covert speech vs. covert humming Schultz et al. 2005: voiced vs. whispered speech

Toyomura et al. 2007: COMP

EXP 1: PUTAMINAL ACTIVITY IGNORE - SIMPLE COMP - SIMPLE Put p. STS Put 4. 0 4. 4 2. 0 z = 10 2. 4 z=2