LH LH LH LH LH LH LH LH LH LH LH LH ... - Semantic Scholar

Sensorimotor Transformation RSA Searchlight Results - Vocal Tract Models LH

z:

-44 LH

z:

-35 LH

-26

-16

LH

z:

LH

LH

-7

LH

12

22

LH

30

LH

z:

50

LH

2 LH

40 p < 0.00001

LH

p < 0.021 (FDR q < 0.05)

58 Vocal tract models individual subjects

Individual Subject Vocal Tract RDMs S01

S07

S13

S19

/i/

/i/

/i/

/i/

/y/

/y/

/y/ /a/

/a/

/a/ /ɶ/

/a/ /ɶ/

/ɶ/

/ɶ/

S02

S08

S14

S20

/i/

/i/

/i/

/i/

/y/

/y/ /a/

/y/ /a/

/ɶ/

/y/ /a/

/ɶ/

/a/ /ɶ/

/ɶ/

S03

S09

S15

S21

/i/

/i/

/i/

/i/

/y/

/y/ /a/

/y/

/y/

/a/ /ɶ/

/a/ /ɶ/

/a/ /ɶ/

/ɶ/

S04

S10

S16

S22

/i/

/i/

/i/

/i/

/y/

/y/ /a/

/y/

/y/

/a/ /ɶ/

/a/ /ɶ/

/a/ /ɶ/

/ɶ/

S05

S11

S17

S23

/i/

/i/

/i/

/i/

/y/

/y/ /a/

/y/

S06

(b)/i/

/a/ /ɶ/

S12 /i/

/y/

/ɶ/

/a/

RH

/a/

/ɶ/

S18

/i/

/i/

/y/ /a/ /ɶ/

/a/

LH

/ɶ/

/a/ /ɶ/

CTM 1

/i/ Vocal tract: group average model/y/

/y/

/y/

/a/ /ɶ/

LH

/y/

/ɶ/

CTM 2

High dissimilarity

/i/ /y/

/y/

CTMs

/a/ /ɶ/

/a/LH

/ɶ/

Low dissimilarity RH

LH

RH

/a/ average model /y/ average model production data;3d testing Fig. average model ɶ stimulus /ɶ/ average

ɶt /models

for effects of training & g Performance LH native/non-native vowel. Least successful LH

/a/

x: 15

z: 0

RHz: -20 RH

RH

RH

LH

Fig. 3d Fig. 3d cal tract: group (a)

y: -60

RH RH y: -5 LH LH

p=

RH

p = 10-3

10-2

RH

LH

z: 15

RH y: -6

y: -1

p = 10-2

p = 10-3

LH

p = RH 10-3

y: RH

(b)

/y/

/a/

LH

-2 p = 10p

=

p = 10-2 -2 10

(b)

p=

CTM 2

x: -53

10-2

/i/

z: 43

RH

RH

x: 50

RH

y: -6

/a/

(b) Univariate fMRI (BOLD respons y: -6

p = 10-2

CTM 1

/i/

/i/ Analysis of speech /i/ /y/ Vocal tract: group /y/ /y/group perception and Vocal Vocal t (b) tract: /a/ /a/ /a/ average model production data; testing average model avera /ɶ/ /ɶ/ /ɶ/ for effects of training & native/non-nativeLHvowel. LH

(b)

/y/ /ɶ/

LH

RH

p = 10-2

rtMRI blocks x: -64 (x3 pairs) fMRI blocks (x3) Structural scan (s)

LH

Low dissimilarity

y: -6p = 10-2

Group avg.

Stimulus avg.

CTM 1

Subjects imitate z: 0 z: 0 2 trained and 2 (20 trials)LH LH10-3 untrained p= p = 10-3 vowels. Stimulus avg.

Group avg.

Least successful

Most Successful

1 trial (10 per block)

High dissimilarity

LH

p = 10-3

-3 Scanning p = 10-2 (b) -2 p = 102. p = 10-3 p = 10 Perception ROIs

p = 10-3

p = 10-2

/ɶ/

Vocal tract: group y: -6 average model

Group avg.

Least successful

Most Successful

(a)

/i/

Stimulus avg.

Sensorimotor Transformation - Listen Pre-imitate RH RH LH LH RH

=RH 10-3 Least successful

x: -56

y: -5

-3

RH

y: -64

dynamics l for all rusion

LH

RH

z: -2 0 p = 10

Searchlight ROIs LH p = 10

p = 10-3

p = 10-2

/ɶ/

LH

Stimulus avg.

RH

y:x: -64 -61 input

y:y:-6 -5

RI (videos)

RH

LH x: 60p = 10-3 RH

RSA searchlight

y: -6

Group avg.

LH

Stimulus model: tokens collapsed x: -56 z: 15

x: -61

Most Successful

/ɶ/

/ɶ/

Low similarity

RH

p

z: 43

/y/ x: 60 CTMs stimulus average /ɶ/ model /a/ Stimulus PSD Model Low dissimilarity / model LHx: -61 /ɶ z: 0 /y/ /ɶ/ /a/

High similarity

60

RH

/a/ /a/

/i/

LH

y: -6

HRH

/y/

LH RH

ct model: subjects, & tokens LHcollapsed

RH

LH

Group avg.

x:RH-60RH x: 60LH

:-20 7

model LH RH

Least successful

/i/

/y/

ht

/ɶ/

/y/CTM 2 High dissimilarity /i/ Vowel acoustics: stimulus average x: -60 /a/ Vowel Acoustics:

Most Successful

/i/

z: -20

RH

x: -61

Vowel acoustics:

CTM 1

verage model

uccessful

/ɶ/

/a/

p = 10-3 Speech perception: ‘Listen only’ Only Speech Perception - Listen Auditory stimuli (spectrograms) /i/

subjects & blocks collapsed

x: 15

/ /

z: 0

RHz: -20

LH

RH