Pupil dilation reflects the authenticity of received nonverbal vocalizations
Abstract
The ability to infer the authenticity of other’s emotional expressions is a social cognitive process taking place in all human interactions. Although the neurocognitive correlates of authenticity recognition have been probed, its potential recruitment of the peripheral autonomic nervous system is not known. In this work, we asked participants to rate the authenticity of authentic and acted laughs and cries, while simultaneously recording their pupil size, taken as proxy of cognitive effort and arousal. We report, for the first time, that acted laughs elicited higher pupil dilation than authentic ones and, reversely, authentic cries elicited higher pupil dilation than acted ones. We tentatively suggest the lack of authenticity in others’ laughs elicits increased pupil dilation through demanding higher cognitive effort; and that, reversely, authenticity in cries increases pupil dilation, through eliciting higher emotional arousal. We also show authentic vocalizations and laughs (i.e. main effects of authenticity and emotion) to be perceived as more authentic, arousing and contagious than acted vocalizations and cries, respectively. In conclusion, we show new evidence that the recognition of emotional authenticity can be manifested at the level of the autonomic nervous system in humans. Notwithstanding, given its novelty, further independent research is warranted to ascertain its psychological meaning.
Full text
1
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s
Pupil dila ion e lec s
he au hen ici y o ecei ed
non e bal ocaliza ions
Gonçalo Cosme1, Ped o J. Rosa2,3, Césa F. Lima3, Vânia Ta a es1,4, Sophie Sco 5,
Sinead Chen6, Thomas D. W. Wilcockson7,8, T e o J. C aw o d7 & Diana P a a1,3,9*
The abili y o in e he au hen ici y o o he ’s emo ional exp essions is a social cogni i e p ocess
aking place in all human in e ac ions. Al hough he neu ocogni i e co ela es o au hen ici y
ecogni ion ha e been p obed, i s po en ial ec ui men o he pe iphe al au onomic ne ous sys em
is no known. In his wo k, we asked pa icipan s o a e he au hen ici y o au hen ic and ac ed
laughs and c ies, while simul aneously eco ding hei pupil size, aken as p oxy o cogni i e e o and
a ousal. We epo , o he i s ime, ha ac ed laughs elici ed highe pupil dila ion han au hen ic
ones and, e e sely, au hen ic c ies elici ed highe pupil dila ion han ac ed ones. We en a i ely
sugges he lack o au hen ici y in o he s’ laughs elici s inc eased pupil dila ion h ough demanding
highe cogni i e e o ; and ha , e e sely, au hen ici y in c ies inc eases pupil dila ion, h ough
elici ing highe emo ional a ousal. We also show au hen ic ocaliza ions and laughs (i.e. main e ec s
o au hen ici y and emo ion) o be pe cei ed as mo e au hen ic, a ousing and con agious han ac ed
ocaliza ions and c ies, espec i ely. In conclusion, we show new e idence ha he ecogni ion o
emo ional au hen ici y can be mani es ed a he le el o he au onomic ne ous sys em in humans.
No wi hs anding, gi en i s no el y, u he independen esea ch is wa an ed o asce ain i s
psychological meaning.
We exp ess emo ions in social in e ac ions o con ey in o ma ion abou ou a ec i e s a es and in en ions,
which is essen ial o communica ion. In u n, we cons an ly e alua e he au hen ici y behind o he s’ emo ional
exp essions, e en in olun a ily. This complex p ocess o cogni i e empa hy (also known as heo y o mind o
men aliza ion), uses pe cep ual and senso imo o cues1, and allows o an adequa e social esponse, om mo e
in ui i e o mo e delibe a e, such as decisions o us o no o us in he o he indi idual, and hus whe he
o coope a e o o compe e2. Gi en ha hese decisions a e i al o social bonding, de ense om agg ession,
and ul ima ely social ne wo k s uc u e, hey ha e been o u mos impo ance o human su i al3. As such hey
may be ha d-wi ed in ou ne ous sys em. Howe e , i is s ill unknown whe he and how he e alua ion o he
au hen ici y o ano he ’s emo ional exp ession engages he au onomic ne ous sys em o he pe son pe cei ing i .
Emo ions can be e ec i ely exp essed ocally wi hou seman ic con en , such as in laugh e and c ying,
uncons ained by linguis ic s uc u es4–6. E en in he absence o a si ua ional con ex , non e bal ocaliza ions
p o ide ele an cues o in e emo ional s a es7, and hei ecogni ion can anscend cul u es8. Non e bal ocali-
za ions can a y in emo ional ca ego y (e.g., amusemen , sadness, ange , ea , su p ise and disgus )9, alence4,
a ousal4, a ilia i e alue (i.e. emo ional con agion)10 and au hen ici y1. C ying, o example, is an in ense emo-
ional exp ession o a nega i e s a e o en accompanied by lac ima ion, which, in a social con ex , is assumed
o ha e he pu pose o elici ing help om lis ene s11, o , in an in e pe sonal con ex , is unde s ood o unc ion
as elie and imp o e mood a e shed ea s12. Con e sely, and as ly mo e s udied in non e bal ocaliza ion
OPEN
1Ins i u o de Bio ísica e Engenha ia Biomédica, Faculdade de Ciências da Uni e sidade de Lisboa, Campo
G ande, 1749-016 Lisboa, Po ugal. 2HEI-LAB: Human-En i onmen In e ac ion Lab, Uni e sidade Lusó ona
de Humanidades e Tecnologias, Campo G ande, 376, 1749-024 Lisboa, Po ugal. 3Ins i u o Uni e si á io de
Lisboa (Isc e-IUL), CIS-Isc e, A enida das Fo ças A madas, 1649-026 Lisboa, Po ugal. 4Faculdade de Medicina
da Uni e sidade de Lisboa, A enida P o esso Egas Moniz MB, 1649-028 Lisboa, Po ugal. 5Ins i u e o Cogni i e
Neu oscience, Uni e si y College London, Alexand a House, 17-19 Queen Squa e, Bloomsbu y, London,
UK. 6Risk Socie y and Policy Resea ch Cen e , Na ional Taiwan Uni e si y, Roose el Rd., Daan Dis ., Taipei 10617,
Taiwan. 7Depa men o Psychology, Lancas e Uni e si y, Lancas e LA1 4YF, UK. 8School o Spo , Exe cise,
and Heal h Sciences, Loughbo ough Uni e si y, Clyde Williams Building, Epinal Way, Loughbo ough LE11 3GE,
UK. 9Depa men o Neu oimaging, Ins i u e o Psychia y, Psychology and Neu oscience, King’s College London,
16 De C espigny Pa k, Cambe well, London SE5 8AF, UK. *email: [email p o ec ed]
2
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
li e a u e, laugh e is an emo ional exp ession o a posi i e s a e and has he ole o p omo ing and main aining
social bonding13. The unde lying emo ion causing he exp ession o c ying o laugh e may a y such ha acous-
ic di e ences ha e been ound o di e en kinds o laugh e (e.g. icklish e sus emo ional)14,15, o example.
Co ec ly ecognizing he au hen ici y o o he s’ laugh e o c ying is a social skill essen ial o a oiding
decep ion2,16. Au hen ici y ecogni ion is he abili y o disce ning be ween an au hen ic (genuine) e sus an ac ed
(delibe a e) emo ional exp ession, o which we use acous ic di e ences in he case o non e bal ocaliza ions15,17.
We ha e shown ha au hen ic laugh e and c ies a e o en mo e highly pi ched, longe in du a ion and ha e di e -
en spec al cha ac e is ics compa ed o hei ac ed a ian s18. Addi ionally, ac oss a ious emo ions, highe and
mo e a iable pi ch, lowe ha monici y, and less egula empo al s uc u e a e he bes p edic o s o au hen ic-
i y judgemen s19. Lis ene s ecognize au hen ici y in laugh e a oughly 70% accu acy (67%17, and, as we ha e
shown, 72%15 and 63%19). Fu he mo e, au hen ic laugh e is a ed as mo e a ousing and mo e posi i e han
ac ed laugh e 15. Gene ally, genuine emo ional exp essions a e p oduced eac i ely while delibe a e exp essions
a e in en ional and con olled o ms o communica ion13. Whe eas au hen ic laugh e is genuine and usually
an immedia e eac ion o a posi i e and su p ising s imulus, ac ed laugh e is associa ed wi h poli e ag eemen
and ( eal o ake) app ecia ion13,17. While au hen ic c ying is also genuine and usually nega i e s imulus-d i en,
ac ed c ying is associa ed wi h (manipula i e) social decep ion16.
We ha e shown, du ing passi e lis ening, he an e io medial p e on al co ex (amPFC) and an e io cingu-
la e co ex o be mo e s ongly engaged o ac ed ( han au hen ic) laugh e 1. This is consis en wi h he iew ha
in e p e ing non-au hen ic s imuli, and sol ing i s ambigui y, is ela i ely mo e cogni i ely demanding20, engag-
ing cogni i e empa hy (i.e. men aliza ion) o a highe deg ee. In addi ion, a linea dec ease in amPFC ac i a ion
as pe cei ed au hen ici y inc eases18, has been ound. To ou knowledge, he elec ophysiological esponse o
non e bal ocaliza ion au hen ici y has no ye been in es iga ed, bu isual s imuli’s au hen ici y (e.g. genuine
s. ambiguous smiles) has been e lec ed in ea ly e en ela ed po en ials (ERPs) componen s’ ampli ude, e.g.
P200, albei only when he ambiguous smiles we e blended wi h ang y eyes, sugges ing i is p ocessed e y ea ly,
and dependen on he salience o he exp ession21. Mo e e idence ha cogni i e s a egies a e equi ed o in e
au hen ici y and ha hese demand a le el o social ma u i y and expe ience, comes om he indings ha adul s’,
bu no child en’s, cogni i e and emo ional empa hy sco es co ela ed wi h au hen ici y disc imina ion o happy
aces22. In addi ion, high emo ional ai empa hy may also help au hen ici y ecogni ion by acili a ing he simu-
la ion o he emo ion i sel ( h ough emo ional con agion, i.e. he p opensi y o esona e wi h o he s’ emo ions),
leading o a s onge emo ional esponse o au hen ic exp essions22. Indeed, we ha e epo ed ha emo ional ai
empa hy, emo ional con agion and au hen ici y ecogni ion in laugh e o be posi i ely associa ed10.
Pupil size is used as a p oxy o bo h a ousal23 and cogni i e e o in emo ion esea ch24,25 and i depends on
au onomic pe iphe al ne ous sys em ac i i y, which may in u n be elici ed by cen al ne ous sys em inpu .
Ac i a ion o he i is dila o muscle26 s ems om a sympa he ic esponse (known in he con ex o he ‘ igh -o -
ligh ’ mode) igge ed by ad enaline elease, and p oduces an enla gemen o he pupil size (i.e. pupil dila ion).
Ac i a ion o he i is sphinc e muscle26 s ems om a pa asympa he ic esponse ( ypical o he ‘ es and diges ’
mode)23,27, and p oduces a educ ion o he pupil size (i.e. pupil cons ic ion). The pupil dila es wi h highe
a ousal elici ed by a s imulus23, hus, emo ionally cha ged ocaliza ions e oke highe pupil dila ion compa ed o
neu al ones (wi h di e ences be ween posi i e and nega i e showing mixed esul s28). In addi ion, he pupil also
dila es wi h cogni i e e o 29,30, and has been associa ed wi h amPFC ac i i y, a key a ea o cogni i e empa hy as
abo emen ioned31. Fu he mo e, pupil mimic y (i.e. synch oniza ion) du ing social in e ac ion is p oposed o be
an emo ional con agion mechanism and an implici o m o social communica ion32. Indeed, pupil size changes
a e elici ed by emo ional con agion32 which in u n acili a es au hen ici y ecogni ion a leas o laugh e 10.
Examining how sensi i e is, i a all, pupil size o he au hen ici y o pe cei ed emo ional exp essions can
e eal unde lying p ocesses o au hen ici y disc imina ion; in pa icula , whe he he au onomic sys em is
in ol ed. Fu he mo e, by combining his examina ion wi h empa hy ai s and beha iou al measu emen s,
one could disen angle o wha ex en au hen ici y disc imina ion is a cogni i ely demanding and/o a ec i e
p ocess. Howe e , he di ec associa ion be ween pupil size and he ecogni ion o au hen ici y has no ye been
examined, o ou knowledge. Al hough au hen ici y ecogni ion is an essen ial cogni i e empa hy skill o an
adap i e social beha io , and i s cen al ne ous sys em co ela es ha e s a ed o be un a eled, i is s ill unknown
whe he i engages he pe iphe ic au onomic ne ous sys em. He ein, we asked o he i s ime, whe he he
au hen ici y o an emo ional exp ession induced an au onomic ne ous sys em esponse du ing i s pe cep ion. In
he p esen s udy, we es ed his by assessing he e ec o non e bal ocaliza ions’ au hen ici y on he pupil size
o he lis ene . We asked pa icipan s o a e he deg ee o au hen ici y in au hen ic and ac ed laughs and c ies,
du ing which we eco ded changes in hei pupil size and, pos e io ly, he deg ee o emo ional con agion and
a ousal o he same s imuli. Las ly, we measu ed he pa icipan s’ cogni i e and emo ional empa hy ai s. Gi en
he unp eceden ed examina ion in o he au onomic ne ous sys em’s ac i i y du ing au hen ic disc imina ion
and he dual-p oxy na u e o pupil esponse, we had wo possible (di ec ional) p edic ions al e na i e o he
null hypo hesis, one assuming a p eponde ance o a ousal, he o he o cogni i e e o : (1) au hen ic ocaliza-
ions would elici highe pupil dila ion compa ed o ac ed, because hey ha e been ound o be mo e a ousing
in gene al19,33, and pupil dila ion inc eases wi h a ousal23; o (2) by he con a y, au hen ic ocaliza ions would
elici lowe pupil dila ion, because au hen ici y disc imina ion, a leas in laugh e , has been ound o dec ease
he engagemen o p e on al cogni i e empa hy- ele an b ain a eas1,18, sugges ing lowe cogni i e demand. As
such, we aimed o disen angle whe he i is a ousal (supposedly highe in au hen ic) o cogni i e load (suppos-
edly highe in ac ed ocaliza ion), du ing au hen ici y ecogni ion, ha engages au onomic ne ous sys em he
mos . Addi ionally, by including an examina ion o he neu al co ela es o c ying o he i s ime, we explo ed
how ou wo p edic ions would depend on emo ion alence (i.e. would di e be ween laughs and c ies); which
is wa an ed gi en ha he social meaning o au hen ici y in each emo ion can be qui e di e en (i.e. a ake laugh
can signi y benign poli eness o sa casm, bu a ake c y can mean cos ly decei ). Complemen a ily, (1) gi en ha
3
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
au hen ici y disc imina ion has been posi i ely co ela ed wi h cogni i e22 and emo ional empa hy10,22, we es
he associa ion o hese empa hy ai s wi h ou beha io al and pupil size measu es; and (2) we also ask whe he
p e ious posi i e associa ions be ween pe cei ed au hen ici y and a ousal18 and, o he i s ime, emo ional
con agion1,10, in laugh e , ex end o c ying.
Ma e ials and me hods
Pa icipan s. All me hods we e ca ied ou in acco dance wi h ele an guidelines and egula ions. Thi y-
eigh indi iduals we e ec ui ed o pa icipa e in he expe imen ia he labo a o y’s ec ui men websi e. Eigh
pa icipan s we e excluded om analysis due o echnical p oblems in da a acquisi ion (i.e. no eye da a eco ded
ac oss session o ask igge miscon igu a ion), and 2 due o inad e en ly uncon olled oom luminosi y, hus
he inal sample consis ed o 28 pa icipan s (13 male and 15 emale) wi h an a e age age o 23.0yea s (SD = 1.38,
anging om 21 o 26yea s old). The inclusion c i e ia we e igh handedness (Edinbu gh Handedness In en-
o y; due o EEG measu es being collec ed in he same s udy)34 and Eu opean Po uguese as i s language.
Fo emale pa icipan s, an addi ional inclusion c i e ion was o be on he ac i e weeks o he con acep i e
pill, as ime o he mens ual cycle has been shown o a ec emo ion ecogni ion ask pe o mance35. Pa ici-
pan s p o ided w i en in o med consen and we e paid o hei pa icipa ion. To asce ain a no mal dis ibu-
ion in e ms o wo king memo y, emo ional s a e, and a men ally heal hy sample, he Posi i e and Nega i e
A ec Schedule (PANAS)36,37 (Posi i e A ec Sco e—M = 43.41, SD = 8.52; Nega i e A ec Sco e—M = 29.95,
SD = 9.80); he Fo wa d and Backwa d Digi Span Tes s o Wo king Memo y Index o he Weschle Adul In el-
ligence Scale—Thi d Edi ion (WAIS—III)38 (WM Index = 19.71, SD = 3.66); and he B ie Symp om In en o y
was adminis e ed (Global Se e i y Index (GSI)—M = 0.66, SD = 0.49), espec i ely.
S imuli. The se o audi o y s imuli comp ised au hen ic and ac ed non e bal ocaliza ions o amusemen —
laugh e —and o sadness—c ying—along wi h neu al ocaliza ions (e.g. he owel ‘ah’ u e ed wi h neu al
in ona ion)39. The neu al ocaliza ions we e included only o compa ison agains c ies and laughs; he e we e
no ac ed-neu al o au hen ic-neu al s imuli, as neu al s imuli a e, by na u e, no a ec i e and hus canno
be au hen ic o ac ed. Au hen ic ocaliza ions we e elici ed by he speake s while wa ching humo ous ideos
(au hen ic laugh e ) o while ecalling uly upse ing e en s (au hen ic c ying), whe eas ac ed laugh e and
c ying we e ac ed unde ull olun a y con ol. We used ocaliza ions we p e iously alida ed a he beha io al
and neu oimaging le el1,40, as ollows in sho . Th ee male and 3 emale speake s eco ded he s imuli in an
anechoic chambe . Fo au hen ic laugh e , YouTube ideoclips which we e p e iously iden i ied by he speake s
as humo ous, we e shown o induce hem o laugh ou loud. Fo au hen ic c ying, speake s we e encou aged o
ecall pe sonal upse ing e en s and/o s a by posing c ying in o de o ansi ion genuine c ying. Las ly, he
speake s we e asked o simula e ac ed laugh e and c ying wi hou eeling any genuine amusemen o sadness,
espec i ely. To a oid ca y-o e e ec s o genuine amusemen o sadness, he eco ding o ac ed laugh e o
c ying always p eceded he eco ding o au hen ic laugh e o c ying. F om each eco ding session, sepa a e
audio iles we e c ea ed o laugh e and c ying, sampled a 44.1kHz o mono.wa iles wi h 16-bi esolu ion. To
con ol o a iabili y in he acous ic p ope ies o he sounds, he audio was no malized o oo -mean-squa e
(RMS) ampli ude using P aa so wa e (www.p aa .o g)40. In his s udy, o each condi ion (au hen ic laugh e ,
ac ed laugh e , au hen ic c ying, ac ed c ying), 18 ocaliza ions we e used and p esen ed wice. An addi ional
60 neu al ocaliza ions we e p esen ed once. In he end, he s imuli se consis ed o 132 ocaliza ions, each
wi h di e en du a ions (in milliseconds—au hen ic laughs: M = 2399.94, SD = 460.73, ange = 1536.00, 3141.00;
ac ed laughs: M = 2248.89, SD = 400.15, ange = 1710.00, 2903.00; au hen ic c ies: M = 2684.55, SD = 289.36,
ange = 2079.00, 2993.00; ac ed c ies: M = 2322.11, SD = 351.48, ange = 1959.00, 2990.00; neu als: M = 2498.74,
SD = 292.08, ange = 2057.00, 2930.00). The acous ic p ope ies o he s imuli (du a ion (ms), mean undamen al
equency—F(0), mean in ensi y (dB)) we e ob ained using P aa so wa e and epo ed in mo e de ail in sup-
plemen al ma e ial (Supplemen a y TableS1).
To compa e he acous ic p ope ies be ween condi ions, he main e ec s o au hen ici y and emo ion we e
es ed using non-pa ame ic K uskal–Wallis H Tes s (due o he no mali y assump ion no being ul illed o
hei mean in each condi ion) and epo ing ε2 o e ec size o du a ion, in ensi y and undamen al equency.
Fu he , pai wise compa isons we e pe o med. The e was an associa ion o au hen ici y wi h pi ch (H(1) = 16.53,
p < 0.001, ε2 = 0.97) and du a ion (H(1) = 9.98, p = 0.002, ε2 = 0.59), whe eby au hen ic ocaliza ions had highe
pi ch and we e longe han ac ed ones; bu no wi h in ensi y (H(1) = 0.28, p = 0.596, ε2 = 0.02). Addi ionally, he e
was an associa ion o emo ion wi h pi ch (H(1) = 63.81, p < 0.001 , ε2 = 3.75), whe eby nega i e ocaliza ions had
highe pi ch han posi i e and neu al, and posi i e mo e han neu al; and in ensi y (H(1) = 62.59, p < 0.001,
ε2 = 3.68) whe e posi i e ocaliza ions had highe in ensi y han nega i e and neu al; bu no wi h du a ion
(H(1) = 2.57, p = 0.227, ε2 = 0.15).
Task. Be o e he s a o he ask, which included concomi an pupil size eco ding (desc ibed below), he
pa icipan s we e in o med ha hey would lis en o sounds, and ha hey would be equi ed o a e he sounds
in e ms o hei pe cei ed au hen ici y. Fo neu al sounds, he pa icipan s we e ins uc ed o jus a end o
he s imulus. Always showing a ixa ion c oss on sc een, a ial s a ed wi h silence o 4000ms plus a ji e o
500ms, ollowed by he p esen a ion o he sound s imuli and hen a 3000ms in e s imulus in e al. A e his, a
7-poin Like scale showed on sc een o up o 5000ms o he pa icipan s o answe hei pe cei ed au hen ic-
i y anging om 1 (“Genuine” —au hen ic) o 7 (“Posed”—ac ed). Fo ease o in e p e a ion and discussion, he
scale was e e sed o he s a is ical analysis. The ask ook 36min o comple e and had 204 ials in a pseudo-
andomized and ixed sequence, balancing condi ion ansi ions ial-by- ial so ha condi ions ansi ioned
equally be ween hemsel es as o minimize he e ec s o pupil habi ua ion41.
4
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
A e he abo e pupillome y- eco ded ask ound, pa icipan s (n = 21 o he 28, due o un o eseen ime
limi a ion) we e ins uc ed o e alua e he pe cei ed a ousal and emo ional con agion o he p e iously p esen ed
ocal s imuli in a 7-poin Like scale (A ousal: 1- Low a ousal, 7- High a ousal; Emo ional con agion: 1- No
con agious a all; 7- Highly con agious), excep o neu al sounds. Di ided in wo blocks, i s hey a ed all 72
emo ional sounds in e ms o hei a ousal (72 sounds in o al, 18 o each condi ion: au hen ic laugh e , ac ed
laugh e , au hen ic c ying, ac ed c ying), and in he second block hey a ed he same sounds o hei con agion.
He ein, a ial consis ed in 1500ms plus 500ms o ji e , hen s imuli p esen a ion ollowed by 1000ms o in e -
s imulus in e al, always wi h a ixa ion c oss on he sc een, a e which he Like scale o a ousal o con agion
(depending o he block) would be shown. This ask las ed o 15min and had 124 ials, also pseudo- andomized
and ixed sequence wi h balanced condi ion ansi ions. He ein, each s imulus was p esen ed once in each block.
Pupil size eco ding. The ixa ion c oss and Like scales we e shown in a Leno o 23.8-inch sc een wi h
1920 × 1080 esolu ion and 60Hz e esh a e. Gaze acking and pupil measu emen s we e eco ded using he
SR Resea ch EyeLink 1000 Plus eye acke . A chin es was used o minimize head mo emen and keep a ixed
dis ance o bo h he sc een and came a, a app oxima ely 56cm o all pa icipan s. Raw da a was collec ed
monocula ly a 1000Hz wi h a e age accu acy o 0.15 isual angle.
A e da a collec ion, he pupil size was down sampled o 250Hz ( o sa e on compu a ional cos s), blinks
and da apoin s 100ms be o e and a e blinks we e conside ed as missing da a. A low-pass il e 4Hz cu -o
equency was applied o he signal. P e- ial baseline was ob ained o each ial as he median pupil size
immedia ely be o e s imuli onse , in an in e al ha was 2% o he whole ial, which a ied depending on he
du a ion o he s imuli (M = 204.14, SD = 4.71ms). This median alue was hen sub ac ed ac oss all da apoin s
o i s ial as ad ised in he li e a u e42. Finally, i he missing da a did no exceed 600ms, as blinks longe han
his a e conside ed mic osleeps43–45, he signal was linea ly in e pola ed46,47. These p ep ocessing s eps we e
employed o all pupil dila ion analyses. Fou ime windows o 1s, a e s imuli onse , we e c ea ed o e alua e
pupil size measu es ac oss ime. As ou s imuli had a iable du a ion, a 4-s analysis pe iod ensu es he inclusion
o peak dila ion and consequen e u n o baseline, so ha bo h peak and mean pupil diame e can be adequa ely
measu ed. The segmen a ion in o ime windows allowed a mo e sensi i e and ho ough assessmen o pupil
esponse o au hen ici y and emo ion, and i s consequen cons ic ion du ing and pos s imuli p esen a ion,
and has been p eceden ly employed48,49. Fo each ial, maximum and mean pupil sizes we e ex ac ed in each
indi idual ime window. Las ly, as ecommended in guidelines o pupillome y p e-p ocessing and analysis50,
and common31,51,52, we ha e excluded om he g oup analysis maximum and mean ou lie pupil size da apoin s
a he le el o he ial. Da apoin s we e conside ed ou lie s i hei mean pe condi ion was abo e o below 1.5
imes he in e qua ile ange ( ollowing s anda d c i e ia53).
P ocedu e. Each pa icipan unde wen he expe imen in one session las ing 2h and a hal , si ing com o -
ably in a quie oom a he Cen e o Clinical Resea ch (Cen o de In es igação Clínica) o he Medical Aca-
demic Cen e o Lisbon (Cen o Académico Médico de Lisboa), whose E hical Commi ee app o ed all expe i-
men al p o ocols. Du ing he ask, he audi o y s imuli was p esen ed binau ally h ough a se o Senheise CX
3.00 ea -canal phones a a com o able lis ening le el ha was indi idually adjus ed a he s a o he expe i-
men . The expe imen was de eloped using MATLAB e sion 8.3.0 (R2014a) wi h Psych oolbox354. Pa icipan s
we e encou aged o espond as in ui i ely as possible. Bu ons o he esponse pad we e ma ked wi h he Like
scale poin s o minimize memo y demands. To acili a e he esponse, pa icipan s we e asked o pu h ee in-
ge s o hei le hand in esponse keys 1, 2 and 3 and ou inge s in he emaining esponse keys. Th ee pauses
o 30s we e dis ibu ed equally along he expe imen o minimize a igue. Concomi an elec oencephalog aphy
eco ding also ook place (da a no ye analyzed). A e he pupillome y- eco ded ask, pa icipan s a ed hei
pe cei ed a ousal and emo ional con agion o e e y sound, and inally esponded o he Empa hy Quo ien
(EQ) and he Reading he Mind in he Eyes Tes (RMET), which assesses emo ional and cogni i e ai empa hy.
S a is ical analysis. Co ela ion analyses (be ween ial-by- ial s imuli and pupillome y measu es;
be ween ai empa hy sco es and pupillome y measu es, wi h au hen ici y as a ca ego ical mode a o ; and
be ween ai empa hy sco es and au hen ici y disc imina ion index—desc ibed below) we e pe o med in R
so wa e 3.655, and when applicable, using he mco package56. To e i y eliabili y o he a ing scales, C on-
bach’s alpha was calcula ed o each one57. To in e he main e ec s o au hen ici y and i s in e ac ion wi h emo-
ion on each beha io al measu e (au hen ici y, a ousal and emo ional con agion a ings), a epea ed measu e
Analysis O Va iance ( pANOVA) model o each measu e was conduc ed in SPSS (IBM Co p. Released 2017,
IBM SPSS S a is ics o Windows, Ve sion 26.0. A monk, NY: IBM Co p.). Fo comple eness, he associa ion o
au hen ici y, a ousal and con agion a ings wi h he wo pupil size measu es was also es ima ed. Fo au hen ici y
disc imina ion, a ings be ween 1 and 3 we e con e ed o ‘posed’, and a ings be ween 5 and 7 we e con e ed
o ‘au hen ic’. Fo comple eness and o a emp a eplica ion o ou p e ious wo k (Ne es e al., 2018), we also
compu ed an index o au hen ici y de ec ion abili y, o each emo ion and each subjec , by sub ac ing he a e -
age au hen ici y a ings o ac ed s imuli om he a e age au hen ici y a ings o au hen ic ones; and we epo ,
in supplemen al ma e ial (Supplemen a y TableS2), he co ela ion be ween hese indexes and indi idual ai
empa hy (EQ and RMET ques ionnai e) sco es.
Pupillome y-wise, wo sepa a e mMANOVA models we e cons uc ed in SPSS, each including Time win-
dow as a wi hin-subjec ac o (0–1, 1–2, 2–3, 3–4s): one o es ima e he main e ec o he wi hin-subjec ac o
Au hen ici y (au hen ic, ac ed) and i s in e ac ion wi h he wi hin-subjec ac o Emo ion (laugh e , c ying); and
ano he o es ima e he main e ec o emo ion (laugh e , neu al, c ying) gi en he inexis ence o au hen ic neu-
al and ac ed neu al sounds. The dependen a iables o bo h mMANOVAs we e he wo pupil size measu es
5
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
(maximum and mean pupil size) o which we epo he wo co esponding uni a ia e mANOVA esul s. All
e ec s o in e es we e ollowed up by pos hoc pai wise compa isons and epo ed a e False Disco e y Ra e
(FDR) co ec ion in R55 (and conside ed s a is ically signi ican when FDR co ec ed p < 0.05). Pa ial E a-squa e
(ηp2) is epo ed as a measu e o e ec size. We epo he main e ec o au hen ici y and emo ion as well as
in e ac ions be ween hem and wi h ime window. Li le’s Missing Comple ely A Random (MCAR)58 es s we e
pe o med o all pupil size measu es o es he andomness o missing alues. Gi en ou p e ious wo k sug-
ges ing he acous ic p ope ies o he s imuli media e au hen ici y ecogni ion19 and since hey na u ally di e
depending on hei au hen ici y and/o emo ion, complemen a y epea ed measu es co ela ion analyses be ween
each s imuli ( ial-by- ial) and pupil size measu es we e conduc ed o e alua e hei di ec e ec on pupil dila ion
and hence hei po en ially media ing ole59 (see Resul s as supplemen a y ma e ial (Supplemen a y TableS5)).
Powe . To ou knowledge, only wo s udies60,61, using a ec i e audi o y s imuli, ha e epo ed e ec s sizes o
he e ec o emo ion on pupil size, whils none ha e epo ed e ec s o au hen ici y. One60 used 26 pa icipan s
o epo a main e ec o emo ion (posi i e, nega i e and neu al) on he mean g adien o pupil diame e , ac oss
0–2s a e s imuli onse , wi h a size o ηp2 = 0.20, while he o he 61 used 97 subjec s o epo a main e ec o
pleasan ness (pleasan , unpleasan and neu al) on pupil diame e wi h a simila e ec size o ηp2 = 0.22 o he
peak ime window (2—4s). We no e hese esul s we e no independen as he la e s udy has a 33% s imuli
o e lap wi h he o me s udy. An a p io i powe analysis, conside ing ηp2 = 0.20, in GPowe 3.1.9.4.62, poin ed
o a need o 22 subjec s o achie e 80% powe , a a 5% alpha in a epea ed measu es ANOVA, o de ec e ec s o
emo ion (3 ca ego ies). We u he aimed ou sample a 38 pa icipan s (and used a inal N = 28), ollowing he
li e a u e’s high s anda d o a ec i e esea ch on pupil size27,28,60,63–65. A sensi i i y analysis indica es ou sample
(N = 28) could de ec (80% powe , 5% alpha): he main e ec o emo ion on pupil dila ion a a minimum e ec
size o ηp2 = 0.16.
Signi icance s a emen . Fo he i s ime, we p obed au hen ici y ecogni ion in human ocaliza ions o
i s e ec on pupil dila ion, a psychophysiological index o men al e o and a ousal. We show ha au hen ic
c ies and ac ed laughs elici ed highe pupil dila ion compa ed o ac ed c ies and au hen ic laughs, espec i ely.
These unp eceden ed indings sugges he socially complex p ocess o au hen ici y ecogni ion in non e bal
ocaliza ions can be e lec ed in a pe iphe al au onomic ne ous sys em esponse, and ha his e ec depends
on he emo ion unde lying he exp ession.
Resul s
Beha io al analysis. Au hen ici y a ing. In e pa icipan eliabili y was high o he au hen ici y a ing
(C onbach’s α = 0.92). Expec edly, au hen ic ocaliza ions (M = 3.39, SD = 0.23) we e pe cei ed as mo e au hen ic
han ac ed (M = 4.46, SD = 0.20) ones [F (1, 27) = 43.84, p < 0.001, ηp2 = 0.62]. The e ec o emo ion was smalle
bu also signi ican , whe eby laughs (M = 3.51, SD = 0.20) we e epo ed as mo e au hen ic han c ies (M = 4.36,
SD = 0.21) [F (1, 27) = 16.61, p < 0.001, ηp2 = 0.38]. No signi ican in e ac ion was ound (p = 0.670).
Pa icipan s ecognized he au hen ici y o laughs and c ies a a le el ha s a is ically signi ican ly exceeded
chance [laugh e : χ2(1) = 145.25, p < 0.001 ; c y: χ2(1) = 152.21, p < 0.001], wi h accu acies o 63.4% and 69.3%
o laugh e and c ying, espec i ely. Au hen ici y disc imina ion indexes we e es ed o co ela ion wi h ai
empa hy sco es, as epo ed in supplemen al ma e ial (Supplemen a y TableS2), wi h no s a is ically signi ican
associa ion ound.
A ousal a ing. In e pa icipan eliabili y was high o he a ousal a ing (α = 0.88). Au hen ic ocaliza ions
(M = 4.10, SD = 0.96) we e pe cei ed as mo e a ousing han ac ed (M = 3.03, SD = 0.80) ones [F (1, 22) = 55.82,
p < 0.001, ηp2 = 0.72]; and laughs (M = 4.13, SD = 0.90) we e pe cei ed as mo e a ousing han c ies (M = 3.00,
SD = 0.95) [F (1,22) = 38.22, p < 0.001, ηp2 = 0.64]. The e was also a signi ican au hen ici y by emo ion in e ac-
ion on a ousal a ings [F (1.22) = 15.05, p = 0.001, ηp2 = 0.43] (whe eby he di e ence wi hin laughs [ (22) = 6.56,
p < 0.001, ηp2 = 0.66], was sligh ly la ge han he di e ence be ween c ies [ (22) = 6.18, p < 0.001, ηp2 = 0.63]).
Emo ional con agion a ing. In e pa icipan eliabili y was also high o emo ional con agion a ing (α = 0.86).
Au hen ic ocaliza ions (M = 4.38, SD = 0.82) we e pe cei ed as mo e con agious han ac ed (M = 3.38, SD = 0.75)
ones [F (1,25) = 70.29, p < 0.001, ηp2 = 0.74]; and laughs (M = 4.21, SD = 0.82) we e pe cei ed as mo e con agious
han c ies (M = 3.50, SD = 0.86) [F (1,25) = 21.70, p < 0.001, ηp2 = 0.47]. No au hen ici y by emo ion in e ac ion
e ec was ound (p > 0.05).
Co ela ions be ween a ings. Repea ed measu es co ela ion analysis indica ed ha he au hen ici y a ing was
posi i ely associa ed wi h he a ousal [ m (1603) = 0.42, p < 0.001], and he con agion a ings [ m (1460) = 0.31,
p < 0.001]; as well as a ousal and con agion a ings be ween hem [ m (1260) = 0.40, p < 0.001].
Pupil size analysis. O e iew. MCAR es s indica ed ha missing alues we e andom o he maximum
and mean pupil sizes o all ime windows (p > 0.999 o all es s). Figu e1 illus a es he pupil diame e o each
condi ion a e aged ac oss all pa icipan s. As he main e ec o au hen ici y on pupil dila ion measu es was no
s a is ically signi ican (p > 0.05), and we e i ied a s a is ically signi ican c oss-o e au hen ici y by emo ion
in e ac ion pa e n, we will only epo and discuss he la e and no he o me .
6
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
Maximum pupil size. The e ec o au hen ici y on maximum pupil size was dependen on emo ion (i.e.
e lec ed in a s a is ically signi ican ‘au hen ici y x emo ion’ in e ac ion) [F (1, 27) = 10.60, FDR co ec ed
p = 0.003, ηp2 = 0.28] and, o a lowe deg ee, on ime window [F (2.32, 62.51) = 4.56, FDR co ec ed p = 0.010,
ηp2 = 0.15]. Pai wise compa isons o he ’emo ion x au hen ici y’ in e ac ion show ha ac ed laughs (M = 216.22,
SD = 12.24) elici ed signi ican ly (FDR co ec ed p = 0.028) highe pupil dila ion han au hen ic ones (M = 192.88,
SD = 10.88), and au hen ic c ies (M = 200.91, SD = 9.92) elici ed signi ican ly (FDR co ec ed p = 0.013) highe
pupil dila ion han ac ed ones (M = 175.87, SD = 11.79), ac oss ime windows. This in e ac ion did no signi i-
can ly di e be ween ime windows, as indica ed by a non-s a is ically signi ican 3-way in e ac ion [F (1.64,
44.12) = 1.59, FDR co ec ed p = 0.332, ηp2 = 0.05]). Ne e heless, o he pu pose o aiding u u e s udies in
selec ing ime windows o pupil size analysis, we epo he ‘au hen ici y x emo ion’ in e ac ion esul s in each
ime window—which we e all s a is ically signi ican (FDR co ec ed p < 0.05)—in he Table1 and in Fig.2.
Tes ed in a sepa a e ANOVA including neu al ocaliza ions, he main e ec o emo ion on maximum pupil
size eached s a is ical signi icance [F (1.56, 42.00) = 21.11, FDR co ec ed p < 0.001, ηp2 = 0.44] as did i s in e ac-
ion wi h ime window [F (2.45, 66.11) = 6.11, FDR co ec ed p = 0.001, ηp2 = 0.19]. Pai wise compa isons show
laughs (M = 210.77, SD = 11.62) elici ed signi ican ly (FDR co ec ed p = 0.015) highe maximum pupil dila-
ion han c ies (M = 191.98, SD = 13.69), and each highe han neu al (M = 149.26, SD = 13.90) (FDR co ec ed
p < 0.001 and p = 0.002, espec i ely). In each ime window, hese compa isons hold s a is ical signi icance excep
in 0–1s, be ween c ies and neu al, and in he 2–3s, be ween laughs and c ies, as shown in Fig.2.
Mean pupil size. The e ec o au hen ici y on mean pupil size signi ican ly depended on emo ion [F (1, 27) = 11.15,
FDR co ec ed p = 0.003, ηp2 = 0.29] and on ime window [F (1.80, 48.54) = 7.12, FDR co ec ed p = 0.005,
ηp2 = 0.21]. As on maximum pupil size, and ac oss ime windows, au hen ic c ies (M = 109.85, SD = 12.61) elici ed
signi ican ly (FDR co ec ed p = 0.013) highe mean pupil dila ion han ac ed ones (M = 85.13, SD = 9.95), bu
unlike o maximum dila ion, he di e ence be ween ac ed laughs (M = 122.68, SD = 9.32) and au hen ic ones
(M = 108.60, SD = 9.92) did no each signi icance (FDR co ec ed p = 0.114). As on maximum pupil size, his
in e ac ion did no di e be ween ime windows, as sugges ed by a non-s a is ically signi ican 3-way in e ac ion
[F (1.76, 47.60) = 1.11, FDR co ec ed p = 0.332, ηp2 = 0.04]). Ne e heless, we epo he ‘au hen ici y x emo ion’
in e ac ion esul s in each ime window—which only eached signi icance (FDR co ec ed p < 0.05) in ime win-
dows 1–2s and 3–4s—in Table1 and in Fig.3.
Tes ed in a sepa a e ANOVA including neu al ocaliza ions, he main e ec o emo ion on mean pupil size
eached s a is ical signi icance [F (1.48, 40.02) = 22.63, FDR co ec ed p < 0.001, ηp2 = 0.46] as did i s in e ac-
ion wi h ime window [F (2.45, 66.21) = 9.28, FDR co ec ed p < 0.001, ηp2 = 0.26]. Like o maximum pupil
size, pai wise compa isons show laughs (M = 117.56, SD = 9.53) elici ed signi ican ly (FDR co ec ed p = 0.006)
highe mean pupil dila ion compa ed o c ies (M = 99.32, SD = 11.00), and each compa ed o neu al (M = 57.69,
SD = 10.41) (FDR co ec ed p < 0.001 and p = 0.001, espec i ely). Also as o maximum pupil size, in each ime
window, he compa isons hold s a is ical signi icance excep in 0–1s be ween c ies and neu al, and in 2–3s
be ween laughs and c ies, as shown in Fig.3.
Associa ion o pupil size wi h ai empa hy sco es. The e was a s a is ically signi ican mode a e nega i e asso-
cia ion be ween mean pupil size and he EQ Cogni i e Empa hy ai sco e in laugh e [ (26) = − 0.42, p = 0.028];
see Table2, bu no c ies (collapsing all ime windows; wi h no s a is ical signi icance eached when conside ing
sepa a e windows). Au hen ici y was no ound o signi ican ly mode a e he associa ion be ween ai sco es
Figu e1. Pupil diame e change ela i e o baseline (in a bi a y uni s (AU)) o each condi ion, a e aged
ac oss all pa icipan s. Dashed e ical lines delinea e he ime windows used o analysis.
7
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
and pupil size measu es in laughs and c ies, i espec i e o ime-window (unco ec ed p > 0.05) (Supplemen a y
TableS3, no wi hin each ime window (da a no shown)).
Beha io al and pupil size analysis. None o he beha io al measu es (au hen ici y, a ousal, emo ional con agion
a ings) we e signi ican ly co ela ed (p > 0.05) wi h pupil size measu es (wi hin all ime windows collapsed)
which is epo ed as supplemen al ma e ial (Supplemen a y TableS4). [This is in eplica ion o an independen
analysis in an independen and equally sized UK sample (da a no shown o published)].
Discussion
We asked whe he he e alua ion o he au hen ici y behind o he s emo ional exp essions migh engage he au o-
nomic ne ous sys em. We did his by in es iga ing whe he he le el o au hen ici y o an emo ional exp ession
a ec s he au onomic ne ous sys em o he pe son du ing pe cep ion. In pa icula , we es ima ed he e ec o
non e bal ocaliza ions’ au hen ici y on he pupil size o he lis ene . Addi ionally, we es ed whe he pupil size
esponse depends on he alence o he emo ion being ocalized—which we ound o be he case. We show ha
a lis ene ’s pupil size is a ec ed by a laugh e ’s and c y’s au hen ici y such ha ac ed laughs induce mo e pupil
dila ion han au hen ic ones, and, e e sely, au hen ic c ies induce mo e pupil dila ion han ac ed ones. These
indings a e no con ounded by in insic na u al di e ences in he acous ic p ope ies o au hen ic and ac ed
ocaliza ions. Addi ionally, we show ha laugh e was pe cei ed as mo e au hen ic, a ousing and emo ionally
con agious han c ying.
In de ail, he in e ac ion be ween au hen ici y and emo ion explained almos one hi d o he a iance in pupil
dila ion (i.e. 28% and 29% o maximum and mean pupil dila ion, espec i ely), le unexplained by he o he
modelled main e ec s and in e ac ions, including ime window. B oken down pe emo ion, laugh e au hen ic-
i y nega i ely explained a la ge po ion (up o 28%) o he a iance unexplained o he wise, wi h he peak e ec
a 0–1s o maximum pupil dila ion, and a 1–2s o mean pupil dila ion. In c ying, au hen ici y posi i ely
explained (i.e. in he opposi e di ec ion o laughs), an equally la ge po ion (up o 28%) o he o he wise unex-
plained a iance wi h he peak e ec a 1–2s o maximum pupil dila ion, and a 3–4s o mean pupil dila ion.
E en hough his ‘au hen ici y x emo ion’ in e ac ion was p esen in all ime windows (a leas o maximum
pupil size), i was mos s a is ically signi ican in he 1–2s ime window ( o bo h pupil measu es, FDR co ec ed
p = 0.002–0.005). In his ime window, mo eo e , he simple au hen ici y e ec s peaked o bo h laugh e and
c ying (and in bo h pupil measu emen s) and we e all s a is ically signi ican (Table1).
Aiming o disen angle whe he i is cogni i e e o (supposedly highe in ac ed ocaliza ions) o a ousal
(supposedly highe in au hen ic ocaliza ions) ha engages he au onomic ne ous sys em he mos du ing
au hen ici y ecogni ion, since bo h ha e been associa ed wi h pupil dila ion27,66, ou esul s seem o suppo
he o me p edic ion o laugh e and he la e o c ying. The cogni i e e o in e p e a ion o he laugh e
Table 1. F-s a is ic (nume a o and denomina o deg ees o eedom), p- alue a e mul iple compa ison
co ec ion wi h False Disco e y Ra e (FDR) and pa ial e a squa ed (ηp2) o he au hen ici y by emo ion
in e ac ion on maximum and mean pupil size, in each ime window, a e p esen ed. Mean di e ence (Mdi ),
s anda d e o di e ence (SDdi ), - es (and deg ee o eedom), p- alue a e FDR co ec ion and ηp2 a e also
p esen ed o he ollow-up pos -hoc pai wise compa isons whe eby au hen ic e sus ac ed ocaliza ions we e
con as ed o each emo ion (whils means and SD pe condi ion a e p o ided in he Resul s’ sec ion ex ).
S a is ical signi icance le el: * p < .05.
Time window
(seconds) Pupil size
measu e
Au hen ici y × emo ion in e ac ion Pai wise compa isons (au hen ic s. ac ed)
F-s a is ic (1, 27) FDR co ec ed
p- alue ηp2Emo ion Mdi SDdi - es (27) FDR co ec ed
p- alue ηp2Signi ican
con as s
0–1
Maximum 5.77 .047* .18 Laugh e − 25.10 41.42 − 3.23 .006* .28 Ac ed > Au hen ic
C ying 0.47 31.38 0.08 .937 < .01 –
Mean 4.12 .052 .13 Laugh e − 12.45 29.11 − 2.26 .064 .16 –
C ying 6.33 28.42 1.18 .249 .05 –
1–2
Maximum 9.36 .005* .26 Laugh e − 27.03 67.86 − 2.11 .044* .14 Ac ed > Au hen ic
C ying 24.98 45.42 2.91 .014* .24 Au hen ic > Ac ed
Mean 13.82 .002* .34 Laugh e − 33.36 55.36 − 3.19 .007* .27 Ac ed > Au hen ic
C ying 18.59 46.18 2.13 .042* .14 Au hen ic > Ac ed
2–3
Maximum 6.61 .032* .20 Laugh e − 24.33 75.27 − 1.71 .099 .10 –
C ying 25.33 56.78 2.36 .051 .17 –
Mean 2.75 .109 .09 Laugh e − 14.91 85.07 − 0.93 .362 .03 –
C ye 21.69 56.28 2.04 .103 .13 –
3–4
Maximum 6.79 .019* .20 Laugh e − 16.90 80.61 − 1.11 .277 .04 –
C ying 49.36 85.42 3.06 .010* .26 Au hen ic > Ac ed
Mean 6.23 .019* .19 Laugh e 4.42 75.92 0.31 .760 < .01 –
C ying 52.26 86.27 3.21 .007* .28 Au hen ic > Ac ed
8
Vol:.(1234567890)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
inding is consis en wi h: (1) ac ed ( s. au hen ic) laughs (and hei subjec i e disc imina ion) ha ing been ound
o inc ease engagemen o p e on al cogni i e empa hy- ele an b ain a eas1,18, sugges ing highe cogni i e
Figu e2. Box plo s o maximum pupil diame e as a unc ion o au hen ici y pe emo ion, showcasing
he au hen ici y by emo ion in e ac ion (abo e), and o maximum pupil dila ion as a unc ion o emo ion
showcasing he main e ec o emo ion (below)—bo h o each ime window (seconds). S a is ically signi ican
(p < .05, a e FDR co ec ion) pai wise compa isons a e highligh ed wi h an *. E o ba s ep esen ± 1.5 SD;
AU = a bi a y uni s.
9
Vol.:(0123456789)
Scien i ic Repo s | (2021) 11:3733 | h ps://doi.o g/10.1038/s41598-021-83070-x
www.na u e.com/scien i ic epo s/
demand; (2) in he p esen s udy, only in laugh e , pupil size was nega i ely co ela ed wi h he cogni i e—
and no he emo ional—empa hy sco e, and (3) au hen ici y disc imina ion in acial s imuli inc easing wi h
Figu e3. Box plo s o mean pupil diame e as a unc ion o au hen ici y pe emo ion, showcasing he
au hen ici y by emo ion in e ac ion (abo e) and o mean pupil dila ion as a unc ion o emo ion showcasing
he main e ec o emo ion (below)—bo h o each ime window (seconds). S a is ically signi ican (p < .05, a e
FDR co ec ion) pai wise compa isons a e highligh ed wi h an *. E o ba s ep esen ± 1.5 SD; AU = A bi a y
Uni s.