Passive exposure to speech sounds modifies change detection brain responses in adults
Full text
This is a sel -a chi ed e sion o an o iginal a icle. This e sion
may di e om he o iginal in pagina ion and ypog aphic de ails.
Au ho (s):
Ti le:
Yea :
Ve sion:
Copy igh :
Righ s:
Righ s u l:
Please ci e he o iginal e sion:
CC BY-NC-ND 4.0
h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
Passi e exposu e o speech sounds modi ies change de ec ion b ain esponses in adul s
© 2018 Else ie Inc.
Accep ed e sion (Final d a )
Ku kela, Ja i; Hämäläinen, Ja mo; Leppänen, Paa o H.T.; Shu, Hua; As ikainen,
Piia
Ku kela, J., Hämäläinen, J., Leppänen, P. H., Shu, H., & As ikainen, P. (2019). Passi e exposu e o
speech sounds modi ies change de ec ion b ain esponses in adul s. Neu oImage, 188, 208-216.
h ps://doi.o g/10.1016/j.neu oimage.2018.12.010
2019
Accep ed Manusc ip
Passi e exposu e o speech sounds modi ies change de ec ion b ain esponses in
adul s
L.O. Ku kela Ja i, A. Hämäläinen Ja mo, H.T. Leppänen Paa o, Shu Hua, As ikainen
Piia
PII: S1053-8119(18)32154-2
DOI: h ps://doi.o g/10.1016/j.neu oimage.2018.12.010
Re e ence: YNIMG 15478
To appea in: Neu oImage
Recei ed Da e: 27 June 2018
Re ised Da e: 3 Decembe 2018
Accep ed Da e: 5 Decembe 2018
Please ci e his a icle as: Ku kela Ja i, L.O., Hämäläinen Ja mo, A., Leppänen Paa o, H.T., Hua, S.,
Piia, A., Passi e exposu e o speech sounds modi ies change de ec ion b ain esponses in adul s,
Neu oImage (2019), doi: h ps://doi.o g/10.1016/j.neu oimage.2018.12.010.
This is a PDF ile o an unedi ed manusc ip ha has been accep ed o publica ion. As a se ice o
ou cus ome s we a e p o iding his ea ly e sion o he manusc ip . The manusc ip will unde go
copyedi ing, ypese ing, and e iew o he esul ing p oo be o e i is published in i s inal o m. Please
no e ha du ing he p oduc ion p ocess e o s may be disco e ed which could a ec he con en , and all
legal disclaime s ha apply o he jou nal pe ain.
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
1
Passi e exposu e o speech sounds modi ies change de ec ion b ain
1
esponses in adul s
2
3
Ku kela Ja i L.O.
1*
, Hämäläinen Ja mo A.
1
, Leppänen Paa o H.T.
1
, Shu Hua
2
, As ikainen
4
Piia
1
5
6
1
Depa men o Psychology, Uni e si y o Jy askyla, Jy äskylä, Finland
7
2
Na ional Key Labo a o y o Cogni i e Neu oscience and Lea ning, Beijing No mal
8
Uni e si y, Beijing, China
9
10
* Co esponding au ho : Depa men o Psychology, Ma ilanniemi 6-8, 40014 Uni e si y
11
o Jy askyla, Jy äskylä, Finland; phone: + 358 408054538; e-mail: ja i.ku kela@jyu. i
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
2
ABSTRACT
33
In ea ly li e audi o y disc imina ion abili y can be enhanced by passi e sound exposu e. In
34
con as , in adul hood passi e exposu e seems o be insu icien o p omo e disc imina ion
35
abili y, bu his has been es ed only wi h a single sho exposu e session in humans. We
36
es ed whe he passi e exposu e o un amilia audi o y s imuli can esul in enhanced
37
co ical disc imina ion abili y and change de ec ion in adul humans, and whe he he
38
possible lea ning e ec gene alizes o di e en s imuli. To add ess hese issues, we
39
exposed adul Finnish pa icipan s o Chinese lexical ones passi ely o 2 h pe day on 4
40
consecu i e days. Beha io al esponses and he b ain’s e en - ela ed po en ials (ERPs)
41
we e measu ed be o e and a e he exposu e o he same s imuli applied in he exposu e
42
phase and o sinusoidal sounds oughly mimicking he equency con ou in speech
43
sounds. Passi e exposu e modula ed he ERPs o speech sound changes in bo h igno e
44
(misma ch nega i i y la ency, P3a ampli ude and P3a la ency) and a end (P3b ampli ude)
45
es condi ions, bu no he beha io al esponses. Fu he mo e, e ec o passi e exposu e
46
ans e ed o he p ocessing o he sinusoidal sounds as indexed by he la ency o he
47
misma ch nega i i y. No co esponding e ec s in he ERPs we e ound in a con ol g oup
48
ha pa icipa ed o he es measu emen s, bu ecei ed no exposu e o he sounds. The
49
esul s show ha passi e exposu e o o eign speech sounds in adul hood can enhance
50
co ical disc imina ion abili y and a en ion o ien a ion owa d changes in speech sounds
51
and ha he lea ning e ec can ans e o non-speech sounds.
52
53
54
Keywo ds: pe cep ual lea ning, speech sounds, passi e exposu e, e en - ela ed po en ials
55
56
57
58
59
60
61
62
63
64
65
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
3
1. In oduc ion
66
67
68
In ea ly in ancy, co ical disc imina ion abili y is enhanced e en by passi e sound
69
exposu e alone (e.g., Cheou e al., 1998; Cheou e al., 2002; Kuhl, 2004; T aino , Lee, &
70
Bosnyak, 2011). In con as , in adul hood passi e sound exposu e in absence o aining
71
seems o be insu icien o a ec he neu al-le el disc imina ion abili y (Nää änen e al.,
72
1993; Sheehan e al., 2005; Elme e al., 2017) o beha io al disc imina ion pe o mance
73
(W igh e al., 2010; 2015). Ins ead, e ec s o ac i e disc imina ion aining ha e been
74
shown in se e al s udies by measu ing he misma ch nega i i y (MMN) (K aus e al.,
75
1995; T emblay e al., 1997; T emblay e al., 1998; Tamminen e al., 2015), P3a (A ienza
76
e al., 2004; U he e al., 2006; Seppänen e al., 2012) and P2 (A ienza e al., 2002; Reinke
77
e al., 2003; Sheehan e al., 2005) componen s o e en - ela ed po en ials (ERPs). These
78
componen s e lec p e-a en i e change de ec ion (MMN) and subsequen a en ion
79
shi ing (P3a) based on a memo y ace o med by he lea ned sound ea u e (Nää änen e
80
al., 2005; Polich, 2007) and sound ea u e encoding and s imulus classi ica ion (P2) ( o a
81
e iew see C owley & Col ain, 2004).
82
E en hough e ec s o passi e exposu e ha e been s udied on b ain esponses
83
ela ed o p e-a en i e change de ec ion, possible e ec s o passi e exposu e on a en i e
84
change de ec ion o sounds ha e no been in es iga ed, i.e. e ec s on N2b and P3b
85
componen s. 1-hou a en i e iden i ica ion aining wi h speech sounds, howe e , showed
86
lea ning- ela ed changes in N2b and P3b (Alain e al., 2010). In ano he s udy,
87
iden i ica ion aining esul ed in only enhanced P3b esponses and no changes in N2b
88
(Ben-Da id e al., 2011). Simila ly, a en i e disc imina ion aining wi h speech sounds
89
esul ed in enhanced P3b-like bu no N2b-like mic os a es in elec oencephalog aphy
90
(Gi oud e al., 2017).
91
E en i p e ious s udies ha e ailed o demons a e e ec o passi e exposu e on
92
audi o y change de ec ion in adul s (Nää änen e al., 1993; Sheehan e al., 2005; Elme e
93
al., 2017; W igh e al., 2010; 2015), passi e exposu e o sounds seems no o be en i ely
94
ine ec i e ei he . Pe cep ual lea ning on an audi o y disc imina ion ask (W igh e al.,
95
2010) o on an iden i ica ion ask (W igh e al., 2015) ha is combined wi h sessions o
96
passi e exposu e is mo e e icien han he ac i e aining alone as indexed by beha io al
97
esponses (W igh e al., 2010; 2015). Fu he mo e, passi e exposu e o sounds inc eases
98
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
4
ampli ude o he P2 componen (Sheehan e al., 2005; T emblay e al., 2007; T emblay e
99
al., 2010; Ross e al., 2013). Thus, passi e exposu e seems o ha e a leas acili a ing
100
e ec on audi o y pe cep ual lea ning in adul hood.
101
One possible eason o he ailu e o he p e ious s udies in demons a ing he e ec
102
o passi e exposu e on disc imina ion abili y can be he sho , 1 - 2 hou , exposu e ime
103
ha has been used in p e ious s udies (Nää änen e al., 1993; Sheehan e al., 2005; Elme
104
e al., 2017). Ac i e aining s udies ha e p o ided aining o e se e al days, and his has
105
led o be e disc imina ion abili y as indexed by he enhancemen o he MMN, P3a and
106
P3b esponses (K aus e al., 1995; T emblay e al., 1997; Gi oud e al., 2017).
107
Fu he mo e, i has been shown ha sleep dep i a ion hinde s he lea ning- ela ed inc ease
108
in he MMN ampli ude and p e en s he appea ance o he P3a componen (A ienza e al.,
109
2004). Thus, he lea ning- ela ed changes in co ical esponses seem o be sleep-
110
dependen , p obably equi ing memo y consolida ion du ing noc u nal sleep (Alain e al.
111
2015). Based on his assump ion, i could be possible ha he e ec s o me e passi e
112
exposu e eme ge i he exposu e is expanded on se e al days, allowing memo y
113
consolida ion. This has no ye been es ed explici ly, howe e .
114
The e idence on gene aliza ion o he audi o y lea ning o s imulus ea u es no
115
encoun e ed du ing aining is sca ce. The e a e some s udies showing ha equency o
116
syllable disc imina ion aining gene alizes o closely simila un ained s imuli ( o a
117
e iew see W igh & Zhang, 2009). One s udy applied MMN o s udy he gene aliza ion,
118
and showed ha ca ego iza ion aining o labial s op consonan gene alizes also o
119
al eola s op consonan as indica ed by he sho ened la ency and inc eased ampli ude o
120
he MMN o non- ained s imuli (T emblay e al., 1997).
121
In he p esen s udy, we es ed wo highly no el aspec s o audi o y pe cep ual
122
lea ning: i) E ec o passi e speech sound exposu e on change de ec ion and a en ion
123
o ien ing in igno e and a end es condi ions, and ii) i he e ec o passi e exposu e is
124
obse ed, whe he i gene alizes o igno ed non-speech s imuli. Adul na i e Finnish
125
pa icipan s we e exposed o speech sounds (changes in Chinese lexical ones) o a o al
126
o 8 hou s o e 4 days. ERPs we e eco ded be o e and a e he exposu e o he same
127
speech sounds and also o sinusoidal sounds oughly mimicking he pi ch con ou s o he
128
speech sounds. A con ol g oup ecei ed no exposu e bu pa icipa ed only in he ERP
129
eco dings a he same ime in e als as he expe imen al g oup.
130
We expec ha he passi e exposu e would esul in modula ions in he ERPs,
131
e lec ing changes in bo h p e-a en i e and a en i e change de ec ion and a en ion
132
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
5
o ien ing owa d changes (MMN, P3a, N2b, and P3b), as he exposu e ime is longe han
133
in he p e ious s udies (Nää änen e al., 1993; Sheehan e al., 2005; Elme e al., 2017) and
134
allows memo y ace consolida ion du ing he nigh s be ween he exposu e pe iods
135
(S ickgold, 2005; Alain e al., 2015). Changes in hese ERP componen s a e assumed o
136
occu due o he o ma ion o long- e m memo y ep esen a ions o he sounds, making
137
change de ec ion and a en ion o ien ing o hem mo e e icien (as in Nää änen e al.,
138
1997; Winkle e al., 1999). We also hypo hesized, based on he indings on sound
139
equency aining (W igh & Zhang, 2009), ha he e ec o passi e exposu e ans e s o
140
he non-speech sounds.
141
142
143
2. Ma e ial and me hods
144
145
146
2.1 Pa icipan s
147
148
A o al o 39 monolingual Finnish-speaking pa icipan s (mean age = 23.0 yea s,
149
s anda d de ia ion [SD] = 3.3 yea s; 32 emales and 7 males) olun ee ed o he s udy.
150
They we e ec ui ed wi h announcemen s in he no ice boa ds and e-mail lis s o he
151
Uni e si y o Jy äskylä. The inclusion c i e ia o he s udy we e an age o 18–30 yea s,
152
igh -handedness, no mal hea ing measu ed using audiome y, and sel - epo ed no mal
153
ision (o co ec ed o no mal ision). The exclusion c i e ia o he s udy we e
154
neu ological o psychia ic diso de s, including sleep p oblems, and exposu e o o aining
155
in onal languages. Howe e , p e ious exposu e du ing ips o coun ies whe e onal
156
languages a e spoken (maximum o 2 weeks) was accep ed. W i en in o med consen was
157
ob ained om each pa icipan be o e inclusion in he s udy. The expe imen was
158
unde aken in acco dance wi h he Decla a ion o Helsinki, and he e hical commi ee o
159
he Uni e si y o Jy äskylä app o ed he esea ch p o ocol.
160
The pa icipan s we e di ided o wo g oups, one g oup o pa icipan s we e
161
passi ely exposed o speech sounds (n = 18, mean age = 21.7 yea s, SD ± 1.7) and he
162
o he se ed as a con ol g oup (n = 21, mean age = 24.1 yea s, SD ± 3.6). Da a was
163
collec ed in igno e and a end es condi ions (desc ibed below). F om bo h igno e and
164
a end condi ions da a o 3 pa icipan s we e omi ed om s a is ical analysis due o
165
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
6
ex ensi e a i ac s in he EEG. A e he omission, da a o 18 and 21 con ol g oup
166
pa icipan s and 18 and 15 passi e exposu e g oup pa icipan s emained o he igno e and
167
a end es condi ions, espec i ely. In he igno e es condi ion da a, 66.6% o he
168
pa icipan s in he exposu e g oup and he same po ion o he pa icipan s in he con ol
169
g oup had some musical aining o had played an ins umen o sang as a hobby. In he
170
a end es condi ion da a, his was he case o 66.6% o he exposu e g oup and 61.9% o
171
he con ol g oup pa icipan s. All he pa icipan s had s udied English and Swedish as a
172
o eign language. In addi ion, in he igno e es condi ion da a, 61.1% o he exposu e
173
g oup and 88.9% o he con ol g oup pa icipan s had s udied an addi ional language o
174
o e 2 yea s. In he a end es condi ion da a, his was he case o 73.3% o he exposu e
175
and 81.0% o he con ol g oup pa icipan s.
176
177
2.2 S imuli
178
179
We exposed he pa icipan s o lexical ones, since Finnish belongs o a quan i a i e
180
language g oup, and onal changes a e no pa o he phonological sys em in his
181
language. The e o e, we expec ed ha aining e ec s could be obse ed. Because
182
disc imina ion h eshold o he lexical ones applied in he s udy was no known o he
183
Finnish pa icipan s, and we did no wan he pa icipan s o ac i ely lis en o he sounds,
184
wo le els (la ge and small) o change we e selec ed o maximize he possibili y o ind an
185
exposu e e ec .
186
The sounds we e p epa ed so ha he i s phoneme /a/ was spoken by a emale
187
na i e Chinese speake wi h ising (i.e., Chinese lexical one 2) and alling (i.e., Chinese
188
lexical one 4) pi ch con ou , and hey we e eco ded a a sampling a e o 44.1 kHz. The
189
sounds we e hen digi ally edi ed using SoundFo ge so wa e (SoundFo ge 9, Sony
190
Co po a ion, Japan) o modi y hem o ha e a du a ion o 200 ms. To isola e he lexical
191
ones and keep he es o he acous ic ea u es iden ical, pi ch ie ans e was pe o med
192
using P aa so wa e (P aa 5.4.06, Uni e si y o Ams e dam). Pi ch ie ans e
193
gene a ed a ising one and a alling one, which we e iden ical o each o he , excep o a
194
pi ch con ou di e ence in undamen al equency (F0). These wo ones we e aken as he
195
endpoin s imuli o c ea e a con inuum o lexical ones wi h 10 in e al s eps. A mo phing
196
echnique was pe o med in MATLAB (Ma hWo ks, Inc., MA, US), and a STRAIGHT
197
ool (Kawaha a e al., 1999) was used o c ea e he h ee ones applied in he expe imen .
198
The epea edly p esen ed s anda d sound was he alling one (Fig. 1A), and de ian
199
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
7
sounds we e a sligh ly alling one (small de ian ), and a ising one (la ge de ian , Fig.
200
1A) co esponding o he ones 11, 7, and 3, espec i ely, on he one con inuum. All
201
s imuli we e no malized o ha e he same oo mean squa e in ensi y. The de ailed
202
p ocedu e conce ning how he s imuli we e gene a ed was epo ed p e iously elsewhe e
203
(Xi e al., 2010).
204
The sinusoidal sounds we e c ea ed using SoundFo ge so wa e (SoundFo ge 9,
205
Sony Co po a ion, Japan) and hey had he same du a ion (200 ms) and s a and end F0 as
206
he co esponding speech sounds. Fo he s anda d sound, he s a ing equency o F0 was
207
312 Hz and i g adually dec eased o 180 Hz. The la ge de ian had he s a ing F0 a 233
208
Hz and i inc eased g adually o 268 Hz (Fig. 1 B). Las ly, he small de ian had a s a ing
209
F0 a 268 Hz and i g adually dec eased o 215 Hz (Fig. 1B).
210
Du ing he p e- and pos -exposu e elec oencephalog am (EEG) eco dings and
211
du ing he exposu e, he sounds we e p esen ed in he oddball condi ion, whe e a
212
equen ly occu ing s anda d s imulus (p obabili y o 0.80) was in e spe sed wi h wo
213
de ian sounds (la ge o small, p obabili y o 0.10 o each), using E-p ime 1.2
214
(Psychology So wa e Tools Inc., Sha psbu g, USA) so wa e, esul ing in 1000 s imulus
215
p esen a ions wi h a sound p essu e le el (SPL) o 70 dB. The in e -s imulus in e al (ISI)
216
a ied andomly be ween 440 and 520 ms (o se o onse ). The s imuli we e deli e ed in a
217
pseudo andom ashion, wi h he es ic ion ha consecu i e de ian sounds we e sepa a ed
218
by a leas wo s anda d sounds.
219
220
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
14
Table 1. Summa y o he signi ican e ec s in he epea ed measu es o ANOVA o he a end condi ion
409
(speech sounds). * ma ks exposu e- ela ed e ec . Deg ees o eedom (d ), F- alues (F), P- alues (P), and
410
pa ie al e a squa ed (η
) o e ec sizes a e epo ed.
411
Componen / a iable
E ec
d
F
P
N2b/
ampli ude
S imulus ype
1,34
37.9
0.0001
0.53
De ian ype
1,34
5.2
0.030
0.13
Session
1,34
57.9
0.0001
0.63
Elec ode x G oup
2,33
3.5
0.043
0.17
De ian ype x S imulus ype
1,34
38.8
0.003
0.23
S imulus ype x Elec ode
2,33
6.4
0.004
0.28
Session x Elec ode 2,33
4.8
0.014
0.22
S imulus ype x elec ode x session 2,33
4.9
0.013
0.23
N2b (de ian )/
la ency
De ian ype
1,34
77.36
0.0001
0.70
Session
1,34
9.46
0.004
0.22
Elec ode
2,33
9.91
0.0001
0.38
De ian ype x Session
1,34
4.17
0.049
0.11
P3b/
ampli ude
De ian ype
1,34
32.18
0.0001
0.47
S imulus ype
1,34
97.03
0.0001
0.74
Session
1,34
4.68
0.038
0.12
Elec ode
2,33
22.08
0.0001
0.57
Elec ode x G oup
2,33
4.26
0.023
0.21
De ian ype x S imulus ype
1,34
34.93
0.0001
0.51
S imulus ype x Session
1,34
5.37
0.027
0.14
S imulus ype x Elec ode
2,33
33.64
0.0001
0.67
Session x Elec ode
2,33
5.05
0.012
0.23
S imulus ype x Elec ode x G oup
2,33
9.36
0.001
0.36
De ian ype x S im ype x Elec ode
2,33
6.42
0.004
0.28
De ian ype x S im ype x Session x G oup*
1,34
6.50
0.015
0.16
P3b (de ian )/
la ency
De ian ype
1,34
8.99
0.005
0.21
Session
1,34
59.89
0.0001
0.64
Elec ode
1,34
4.08
0.027
0.20
412
3.1.4 Co ela ions be ween ERPs in a end condi ion and beha io al esponses
413
414
In he exposu e g oup, he e was a ma ginally signi ican co ela ion be ween he
415
pos measu emen P3b ampli ude and eac ion imes o he small de ian , = –0.497, p =
416
0.059, 99% CI [–0.80, –0.11]; he la ge he esponse ampli ude was, he as e he eac ion
417
ime became. O he co ela ions we e non-signi ican .
418
419
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
15
420
421
Fig. 2. G and a e aged P3b esponses in he exposu e g oup, n =15 and he con ol g oup, n = 21. The
422
g ay lines ep esen esponses o de ian sounds om he p e measu emen , and he black lines signi y he
423
esponses o de ian sounds om he pos measu emen . G and a e aged wa e o ms a e p esen ed as mean
424
alues om he collapsed elec ode clus e s (le , middle, and igh pa ie al; see Supplemen a y Fig. 1 and 3).
425
In lowe panel, g and a e aged scalp opog aphies o he de ian esponses as a mean ampli ude alue o he
426
analyzed ime window a 360–410 ms om he 128 elec odes o he P3b o he exposu e g oup and con ol
427
g oup a e shown
.
428
429
430
431
432
433
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
16
434
435
Fig. 3. Passi e exposu e enhanced he P3b ampli ude. The mean ampli ude alues o P3b a e aged ac oss
436
he h ee elec ode clus e s (le , middle, and igh pa ie al) in 360–410 ms ime window. The whi e ba s
437
indica e he ampli ude alues a he p e measu emen , and he g ay ba s ep esen hose a he pos
438
measu emen . *** indica es a s a is ically signi ican di e ence (p < 0.05) and e o ba s indica e he
439
s anda d e o o he mean. Pos exposu e esponses o he de ian sounds we e enhanced compa ed o hose
440
in p e measu emen . No such e ec was ound in he con ol g oup.
441
442
443
3.2 Igno e condi ion o speech sounds
444
445
3.2.1 MMN componen
446
447
The e was no exposu e e ec o he ampli ude o he MMN esponse. Howe e , o
448
he MMM la ency, an in e ac ion e ec o session x g oup was ound (Table 2). Sepa a e -
449
es s o he g oups compa ing he la encies be ween p e and pos measu emen s showed
450
ha he de ian esponse la encies became sho e om he p e o pos measu emen s in
451
he exposu e g oup, while no such changes in la encies we e ound in he con ol g oup
452
(Fig. 4 and 5).
453
454
3.2.2 P3a componen
455
456
Fo he ampli ude o he P3a componen , an in e ac ion e ec o s imulus ype x
457
session x g oup was ound (Table 2). Subsequen ANOVA (s imulus ype x session)
458
pe o med sepa a ely o each g oup e ealed ha he e was an in e ac ion e ec o
459
s imulus ype x session in he exposu e g oup, F
1,17
= 5.66, p = 0.029, η
= 0.25, while no
460
session- ela ed main o in e ac ion e ec s we e obse ed in he con ol g oup
461
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
17
(Supplemen a y Table 4). Sepa a e - es s o s anda d and de ian esponses compa ing
462
he ampli ude change om p e o pos measu emen we e conduc ed o he exposu e
463
g oup. Responses o de ian sounds inc eased in ampli ude owa d a posi i e pola i y and
464
he same was obse ed o he esponses o he s anda d sounds (Fig. 4 and 5).
465
The passi e exposu e a ec ed he la encies o he de ian esponses in he P3a ime
466
window, as indica ed by he de ian ype x session x g oup in e ac ion e ec (Table 2).
467
Subsequen ANOVA (de ian ype x session) pe o med sepa a ely o each g oup
468
e ealed an in e ac ion e ec o de ian ype x session in he exposu e g oup, F
1,17
= 9.01,
469
p = 0.008, η
= 0.35. No session- ela ed main o in e ac ion e ec s we e ound in he
470
con ol g oup. Subsequen - es s compa ing la encies sepa a ely o he de ian ypes in
471
he exposu e g oup showed ha a e he exposu e, he la ency o P3a o he la ge change
472
was sho e (273.6 ms ± 14.9) han i was be o e he exposu e (294.0 ms ± 24.5), (17) =
473
2.92, p = 0.02, 95% CI [8.0, 34.0], d = 1.00 (Fig. 5). No e ec was ound o he small
474
change (Fig. 5).
475
476
Table 2. Summa y o he signi ican e ec s in he epea ed measu es o ANOVA o he igno e condi ion
477
whe e he speech sounds we e p esen ed. * ma ks exposu e- ela ed e ec . Deg ees o eedom (d ), F- alues
478
(F), P- alues (P), and pa ie al e a squa ed (
) o e ec sizes a e epo ed.
479
480
Componen / a iable
E ec
d
F
P
MMN/
ampli ude
S imulus ype
1,34
6.73
0.014
0.17
MMN (de ian )/
la ency
Elec ode
2,33
5.14
0.011
0.24
De ian ype
1,34
6.92
0.013
0.17
De ian ype x Session
1,34
4.17
0.049
0.11
Session x G oup*
1,34
6,71
0.014
0.17
P3a/
ampli ude
De ian ype
1,34
4,27
0.047
0.11
S imulus ype
1,34
4.34
0.045
0.11
Session
1,34
6.91
0.013
0.17
Elec ode
2,33
4.59
0.017
0.22
S imulus ype x G oup
1,34
46.57
< 0.0001
0.58
Session x G oup*
1,34
21.12
< 0.0001
0.38
De ian ype x G oup
1,34
13.76
0.001
0.29
De ype x S im ype x G oup
1,34
12.83
0.001
0.27
S imulus ype x Session x G oup*
1,34
9.90
0.003
0.23
S imulus ype x Elec ode x G oup
2,33
4.95
0.013
0.23
P3a (de ian )/
la ency
Session x G oup*
1,34
4.78
0.036
0.12
De ian ype x Session x G oup*
1,34
6.64
0.014
0.16
481
482
483
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
18
484
Fig. 4. G and a e aged de ian esponses in he ime windows o he misma ch nega i i y (MMN) and
485
P3a esponses in he exposu e g oup, n = 18, and in he con ol g oup, n = 18. G and a e age wa e o ms
486
o de ian esponses (small and la ge de ian a e aged) a e p esen ed as mean alues om he collapsed
487
elec ode clus e s (le on al, middle on al, igh on al; see Ma e ial and me hods and Supplemen a y
488
Fig. 1 and 4). The g ay lines ep esen esponses in he p e measu emen , while black lines ep esen
489
esponses in he pos measu emen o he exposu e g oup and he con ol g oup. The ligh g ay ba s
490
ep esen he ime windows analyzed o he MMN and P3a esponses (190–240 ms and 250–300 ms pos
491
s imulus onse , espec i ely). In lowe panel, he g and a e aged scalp opog aphies o he esponses o
492
de ian sounds as a mean ampli ude alue o he analyzed ime window om he 128 elec odes o he
493
MMN and P3a o he exposu e g oup and he con ol g oup a e shown.
494
495
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
19
496
497
Fig. 5. Misma ch nega i i y la ency and P3a la ency and ampli ude a e enhanced a e he passi e
498
exposu e. The whi e and g ay ba s ep esen he mean alues om he p e and pos measu emen s,
499
espec i ely. E o ba s indica e he s anda d e o o he mean. *** indica es a s a is ically signi ican (p
500
< .05) di e ence. The la ency o he esponse o he de ian sounds in he MMN ime window sho ened
501
om he p e measu emen (222.45 ms ± 19.61)) o he pos measu emen (194.57 ms ± 24.81), (17) = 4.79,
502
p = 0.00, 95% CI [17.44, 39.78], d = 1.25), in he exposu e g oup, while he e we e no changes in he
503
con ol g oup. In addi ion, in he P3a ime window, de ian esponse la encies became signi ican ly sho e
504
o he la ge de ian a e he exposu e, bu no o he small de ian , and he e we e no changes in he
505
con ol g oup. S anda d esponse ampli udes we e mo e posi i e a e he exposu e (–0.05 µV ± 0.64)
506
compa ed o he p e measu emen (–0.53 µV ± 0.55), 17 = 3.18, p = 0.026, 95% CI [–0.79, –0.19], d = 0.80.
507
Also de ian esponse ampli udes became mo e posi i e a e he exposu e (0.57 µV ± 0.57) compa ed o he
508
p e measu emen (–0.30 µV ± 0.74), (17) = 4.70, p = 0.002, 95% CI [–1.23, –0.53], d = 1.32.
509
510
511
512
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
20
3.3 Igno e condi ion o sinusoidal sounds
513
The ans e e ec o non-exposed sound ea u es was es ed by p esen ing
514
sinusoidal sounds oughly mimicking he pi ch con ou s o he speech sounds in a passi e
515
oddball condi ion (Fig. 1C). The ans e e ec was in es iga ed o he componen s and
516
a iables showing g oup x session in e ac ion e ec s in he igno e condi ion whe e speech
517
sounds we e p esen ed, i.e. o he MMN la ency, P3a ampli ude and P3a la ency.
518
3.3.1 MMN componen
519
Fo he MMN la ency, he e was a signi ican in e ac ion e ec o elec ode clus e x
520
session x g oup (Table 3). Subsequen ANOVA (elec ode clus e x session) pe o med
521
sepa a ely o each g oup e ealed an signi ican main e ec session o bo h g oups F
1,17
=
522
15.03, p = 0.001, η
= 0.47; F
1,17
= 5.90, p = 0.027, η
= 0.26 (Supplemen a y able 5).
523
Pos hoc pai ed samples - es s compa ing la encies be ween p e and pos measu emen s
524
sepa a ely o g oups showed ha in he exposu e g oup he la encies go signi ican ly
525
sho e om p e measu emen (234.9 ms ± 10.58) o pos measu emen (215.27 ms ±
526
21.44), (17) = 3.88, p = 0.014, 95% CI [10.43, 29.74], d = 1.16). The e we e no changes
527
in he la encies in he con ol g oup.
528
Table 3. Summa y o he signi ican e ec s in he epea ed measu es o ANOVA o he sinusoidal sounds in
529
he igno e condi ion. * ma ks exposu e- ela ed e ec . Deg ees o eedom (d ), F- alues (F), P- alues (P),
530
and Pa ial e a squa ed (η
) o e ec size a e epo ed.
531
Componen / a iable
E ec
d
F
P
MMN (de ian )/
La ency
Elec ode
2,33
6.89
0.003
0.29
Session x G oup*
1,34
19.57
< 0.001
0.37
Session x Elec ode x G oup*
2,33
4.52
0.018
0.22
P3a/
S imulus
1,34
37.9
< 0.001
0.53
Ampli ude
Elec ode
2,33
8.4
0.001
0.34
S imulus x G oup
1,34
12.4
0.001
0.27
S imulus x Session
1,34
4.2
0.047
0.11
S imulus x Session x Elec ode x G oup*
2.33
4.0
0.027
0.20
P3a (de ian )/
la ency
De ian ype
1,34
7.36
0.01
0.18
De ian ype x G oup
1,34
5.65
0.023
0.14
532
3.3.2 P3a componen
533
Fo he P3a ampli ude, he e was an in e ac ion e ec o s imulus ype x elec ode
534
clus e x session x g oup (Table 3). The ollowing ANOVAs (elec ode clus e x session x
535
g oup) o he de ian esponses (Supplemen a y able 6) o (elec ode clus e x session x
536
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
21
g oup) pe o med sepa a ely o each elec ode clus e e ealed no session x g oup
537
in e ac ion e ec s (Supplemen a y able 7) Subsequen ANOVA (s imulus ype x elec ode
538
clus e x session) pe o med sepa a ely o each g oup e ealed signi ican in e ac ion
539
e ec o s imulus ype x elec ode clus e x session in he con ol g oup, while he e was
540
no session- ela ed e ec s in he exposu e g oup (Supplemen a y Table 8). T- es s
541
compa ing de ian and s anda d esponses sepa a ely om p e measu emen o pos
542
measu emen in each elec ode clus e was pe o med in he con ol g oup. They did no
543
e eal any signi ican esul s.
544
545
546
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
22
547
Fig. 6. G and a e aged MMN and P3a esponses o sinusoidal sounds. G and a e aged wa e o ms o
548
esponses e lec ing MMN and P3a in he exposu e g oup (n = 18) and he con ol g oup (n = 18). The g ay
549
lines ep esen esponses o de ian sounds (la ge and small de ian ypes a e aged) om he p e
550
measu emen , while black lines ep esen esponses o de ian sounds om he pos measu emen o he
551
exposu e g oup and he con ol g oup. The ligh g ay ba s ep esen he ime windows applied in he analysis
552
o o he P3a ampli ude (mean ampli ude alue be ween 300–350 ms). The mean scalp opog aphies o he
553
di e en ial esponse (s anda d sub ac ed om he de ian , de ian ypes a e aged) om he 128 elec odes
554
o he P3a o he exposu e g oup and he con ol g oup om he analyzed ime window. Al hough he e
555
was a signi ican s imulus ype x elec ode clus e x session x g oup e ec o he P3a ampli ude, pos -hoc
556
es s did no e eal any gene aliza ion o he exposu e e ec . Please no e ha he MMN ampli ude was no
557
in es iga ed since he e was no exposu e e ec o i .
558
559
MANUS CRIP T
ACCEP TED
ACCEPTED MANUSCRIPT
23
4. Discussion
560
561
562
He e we show in adul humans ha passi e exposu e o o eign speech sounds o 4
563
consecu i e days, 2 h pe day, enhanced he neu al disc imina ion abili y and a en ion
564
o ien a ion owa d changes in he speech sounds as indexed by ERPs eco ded in igno e
565
and a end es condi ions. The e ec o passi e exposu e o audi o y change de ec ion
566
mechanism has ea lie been ound only in in an s (Cheou e al., 1998; Cheou e al., 2002;
567
Kuhl, 2004; T aino , Lee, & Bosnyak, 2011). In he a end es condi ion, e ec o passi e
568
exposu e was demons a ed as enhanced P3b ampli ude. In he igno e es condi ion,
569
e ec s o passi e exposu e we e demons a ed as sho ened la ency o he MMN and
570
enhanced ampli ude and sho ened la ency o he P3a.
571
The lea ning e ec gene alized o some ex end o no el sounds: he la ency o he
572
MMN sho ened o he sinusoidal sounds no encoun e ed du ing he exposu e phase. This
573
e ec was demons a ed only in he exposu e g oup, no in he con ol g oup.
574
E ec s o audi o y pe cep ual lea ning ha e a ely been es ed o a en i e change
575
de ec ion. He e we showed ha passi e exposu e enhanced he ampli ude o he P3b and
576
he e was end owa ds signi ican co ela ion be ween he enhanced P3b ampli ude and
577
sho ened beha io al eac ion imes o small de ian . P e iously, i has been shown ha
578
when he pe cep ual ask becomes easie , he P3b ampli ude inc eases (Is eal e al., 1985).
579
Ou esul s a e also in line wi h one p e ious s udy which showed ha ac i e aining o
580
disc imina e speech sounds enhances he mic os a es ela ed o he P3b componen
581
accompanied by imp o emen s in beha io al eac ion imes (Gi oud e al., 2017). In he
582
ligh o he con ex -upda ing model (Polich, 2007), passi e exposu e seems o ease
583
compa ison p ocess be ween he ep esen a ion o he s anda d sound in memo y and he
584
de ian sound inpu , which is also e lec ed as sho ened eac ion imes.
585
N2b was he o he componen ha was in es iga ed in he a end es condi ion.
586
He e, he ampli ude o he N2b was no enhanced, no was i s la ency sho ened due o
587
passi e exposu e. In p io s udies applying a en i e aining, in line wi h ou indings,
588
N2b was no enhanced du ing iden i ica ion ask (Ben-Da id e al. 2011) o du ing
589
disc imina ion ask (Gi oud e al. 2017). Howe e , a s udy ha had longe p ac ice pe iod
590
han in s udy by Ben-Da id e al. (2011) epo ed ha enhancemen in abili y o iden i y
591
speech sounds we e ollowed by inc eased N2b (Alain e al. 2010). I emains hus unclea
592