RESEARCH ARTICLE
Se en-mon h-old in an s de ec symme ical
s uc u es in mul i- ea u ed abs ac isual
pa e ns
I ene de la C uz-Pa ı
´aID
1,2,3
*, Gesche Wes phal-Fi ch
4
, W. Tecumseh Fi ch
4
,
Judi Ge ain
3,5
1Depa men o Linguis ics and Basque S udies, Uni e sidad del Paı
´s Vasco/Euskal He iko Unibe si a ea
(UPV/EHU), Vi o ia-Gas eiz, Spain, 2Basque Founda ion o Science Ike basque, Bilbao, Spain,
3In eg a i e Neu oscience and Cogni ion Cen e , CNRS, Uni e si e
´Pa is Ci e
´, Pa is, F ance, 4Uni e si a
¨
Wien, Vienna, Aus ia, 5Depa men o De elopmen al and Social Psychology, Uni e si àdi Pado a,
Pado a, I aly
*i ene.delac [email protected]
Abs ac
The p esen s udy in es iga ed 7-mon h-old in an s’ abili y o pe cei e s uc u al symme y
in mosaic-like abs ac isual pa e ns. We examined in an s’ (n = 98) spon aneous looking
beha iou o mosaic-like sequences wi h symme ical and asymme ical s uc u es.
Sequences we e composed o squa e iles om wo ca ego ies ha di e ed in hei colou
scheme and in e nal shape. We manipula ed sequence leng h (3 o 5 iles) and abs ac ness
o he symme y ( oken s. ca ego y le el). The 7-mon h-olds disc imina ed s uc u ally sym-
me ical om asymme ical mosaics in he i s hal o he es phase ( i s 8 ials). Sequence
leng h, le el o symme y, o numbe o unique iles pe sequence did no signi ican ly modu-
la e in an s’ looking beha iou . These esul s sugges ha e y young in an s de ec di e -
ences in s uc u al symme y in mul i- ea u ed isual pa e ns.
In oduc ion
The oldes human ma king ound o da e is an abs ac zigzag pa e n eng a ed on a shell, c e-
a ed by an ea ly hominin, Homo e ec us, hal a million yea s ago in Ja a, Indonesia [1]. The
ea lies known d awing om ou own species, Homo sapiens, is also abs ac : a c issc oss pa -
e n eng a ed on och e a ound 73,000 yea s ago om Blombos ca e, Sou h A ica [2]. This
abs ac d awing p eda es by abou 30,000 yea s he ea lies known igu a i e pain ing, a hun -
ing scene disco e ed in a ca e in Sulawesi, Indonesia [3]. These indings show ha , om ou
ea lies beginnings, humans ha e p oduced pa e ned abs ac designs. Such designs can be
ound ac oss cul u es, ages, and media: in he Gi ih pa e ns used in Islamic a and a chi ec-
u e, in he ex iles wo en by he Incas, in he deco a ion o Cel ic jewelle y, in Chaco Canyon’s
ce amics, in Maasai shields, o in mode n quil , wallpape , o ab ic designs.
Abs ac isual pa e ns ypically consis o basic uni s ha a e epea ed and/o combined,
and hei a angemen in he plane can o en be desc ibed by a se o ules o , in o he wo ds, a
PLOS ONE
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 1 / 14
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Ci a ion: de la C uz-Pa ı
´a I, Wes phal-Fi ch G, Fi ch
WT, Ge ain J (2022) Se en-mon h-old in an s
de ec symme ical s uc u es in mul i- ea u ed
abs ac isual pa e ns. PLoS ONE 17(5):
e0266938. h ps://doi.o g/10.1371/jou nal.
pone.0266938
Edi o : Claudia Ma¨nnel, Max-Planck-Ins i u u
Kogni ions- und Neu owissenscha en, GERMANY
Recei ed: June 4, 2021
Accep ed: Ma ch 30, 2022
Published: May 11, 2022
Copy igh : ©2022 de la C uz-Pa ı
´a e al. This is an
open access a icle dis ibu ed unde he e ms o
he C ea i e Commons A ibu ion License, which
pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided he o iginal
au ho and sou ce a e c edi ed.
Da a A ailabili y S a emen : All ele an da a a e
wi hin he pape and i s Suppo ing In o ma ion
iles.
Funding: This wo k was suppo ed by he Agence
Na ionale de la Reche che (ANR): G an
[SpeechCode—ANR-15-CE37-0009-01] o JG; he
ANR’s F ench In es issemen s d’A eni – Labex
EFL P og am unde G an [ANR-10-LABX-0083] o
JG; he Eu opean Resea ch Council [Consolida o
G an 773202 ERC-2017-COG ‘BabyRhy hm’] o
JG; he Basque Founda ion o Science Ike basque,
isual “g amma ” [4,5]. As a esul o he s uc u ed combina ion o hei elemen s, hese
isual designs a e o en symme ical.
Symme y o a ious so s is ubiqui ous in he wo ld, cha ac e izing objec s, igu es, and
pa e ns ha occu in na u e (e.g. in physics, biology, chemis y) as well as in all ields o
human c ea ion (e.g. in music, a , poe y) [4]. Adul s appea a uned o pe cei e symme y in
isual s imuli [6–8]: we de ec and disc imina e symme y apidly [9,10], and emembe sym-
me ical displays be e han asymme ical ones [11,12]. Symme y o backg ound elemen s in
a isual sea ch ask acili a es pa icipan s’ iden i ica ion o a a ge , showing ha symme y is
p ocessed au oma ically [13]. Finally, his isual p ope y s ongly impac s ou aes he ic judg-
men s [14,15] and is o en linked o beau y, o ins ance in he case o aces [16].
Adul s de ec h ee main ypes o symme y: mi o , ansla ional, and o a ional symme y.
In mi o symme y, hal o he design is p ojec ed on o he o he hal , as i e lec ed on a mi -
o . In ansla ional symme y, he design is ansposed— epea ed wi hou mi o ing—one o
mo e imes along an axis, while in o a ional symme y a pa e n is o a ed on i s own axis.
Howe e , hese symme y ypes a e no equally salien o he human isual sys em (see [7] o
a e iew): humans appea o be especially sensi i e o mi o symme y, pa icula ly along he
e ical axis [17,18].
The human p e e ence o symme y is also appa en in ou c ea i e p oduc ions. When
Szilagyi & Bai d [19] asked pa icipan s o a ange squa e i ems in o “ isually pleasing” one-,
wo-, o h ee-dimensional displays, hey obse ed ha pa icipan s c ea ed mos ly symme i-
cal designs. In a simila ein, Wes phal-Fi ch and colleagues [20] p esen ed pa icipan s wi h
images o mosaic-like iles o de ed andomly and asked hem o ea ange he a ay o hei
liking, wi hou u he ins uc ion. The majo i y o he esul ing pa e ns we e highly o de ed,
and o e 70% o hem symme ical, including mi o , o a ional and ansla ional symme ies.
Adul s hus pe cei e and p oduce symme y spon aneously, e en when no p omp ed o do
so.
No only is symme y a salien isual p ope y, bu i also helps he isual sys em o ecog-
nise objec s [21,22], seg ega e igu es om he backg ound [23], and i impac s isual sea ch
e iciency [13]. Thus, a e a single sho iew o a no el h ee-dimensional objec , pa icipan s
ecognise he new objec —p esen ed o a ed—signi ican ly be e i i is bila e ally symme i-
cal a he han asymme ical [22]. Simila ly, pa icipan s de ec symme ical wo-dimensional
shapes embedded in a noisy backg ound (i.e. a ays o Gabo elemen s) signi ican ly be e
han asymme ical shapes [23]. Symme y ela ions amongs he elemen s in a scene a e p o-
cessed in pa allel and can acili a e o slow isual sea ch e iciency: i is ha de o de ec a e i-
cally symme ic a ge when i is p esen ed wi h dis ac o s ha a e also symme ical along he
e ical axis, as compa ed wi h dis ac o s symme ical along an oblique axis [13].
The de elopmen al o igins o humans’ pe cep ion o symme y a e no well unde s ood. To
da e, only a hand ul o s udies ha e explo ed human symme y pe cep ion in ea ly in ancy,
using simple shapes, a ays o do s, o pa e ns, which we e always monoch oma ic. The pe -
cep ion o symme y in mo e complex mul i- ea u ed s imuli has no been in es iga ed so a
o young in an s.
The ew a ailable s udies epo ha a 4 mon hs o age— he ea lies age es ed o da e—
in an s disc imina e be ween asymme ical shapes and shapes wi h e ical mi o symme y,
i.e. in which he le hal o he design is mi o ed on o he igh hal [24]. They also look longe
o a ays o do s a anged in e ical mi o symme y p esen ed side-by-side wi h asymme i-
cal o ho izon ally symme ical a ays [25], and habi ua e as e o a simple isual pa e n sym-
me ical along he e ical axis as compa ed wi h simila pa e ns a anged along he
ho izon al o oblique axes, o asymme ically [26,27]. Al hough 4-mon h-old in an s disc imi-
na e be ween shapes wi h e ical and ho izon al symme ies, hey seem o ail o dis inguish
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 2 / 14
o IdlCP; he Spanish Minis y o Science and
Inno a ion [G an n . PID2019-105100RJ-I00] o
IdlCP; he Aus ian Science Fund [FWF He ha
Fi nbe g G an T827-B27] o GWF and FWF DK
G an #W1262-B29] o WTF. The unde s had no
ole in s udy design, da a collec ion and analysis,
decision o publish, o p epa a ion o he
manusc ip .
Compe ing in e es s: The au ho s ha e decla ed
ha no compe ing in e es s exis .
ho izon ally symme ical and asymme ical shapes [24]. These s udies hus sugges a p ocess-
ing ad an age o e ical bila e al symme y, simila o ha a es ed in adul hood [17,18].
The salience o e ical symme y migh be pa icula ly acu e in mi o symme y, since
in an s habi ua e as e o e ical mi o symme y han o e ical ansla ional symme y
[27]. I emains o be es ed whe he in an s disc imina e be ween asymme ical designs and
designs wi h e ical ansla ional symme y. These pionee ing s udies indica e ha an abili y
o de ec symme y in simple, monoch oma ic designs appea s o be p esen in ea ly in ancy.
Impo an ly, in an s do no na iga e a isual wo ld composed o simple one-dimensional
s imuli. Ins ead, hey ace a complex en i onmen in which s imuli con ain mul iple ea u es
(e.g. colou , shape, e c.) ha co-occu in space. And ye , despi e he well-es ablished ole o
symme y in adul isual pe cep ion, i emains unexamined whe he in an s can de ec sym-
me y in mul i- ea u ed s imuli. Mo eo e , all p e ious s udies p esen ed in an s wi h shapes
and a ays ha had pe ec su ace symme y. Whe he in an s a e able o disc imina e sym-
me ical s uc u es lacking pe ec su ace symme y om asymme ical ones is howe e
unknown. While p o iding in an s wi h pe ec ly mi o ed symme ical images allows in an s
o disc imina e symme ical and asymme ical images by elying on low-le el isual mecha-
nisms, p esen ing hem wi h s uc u ally symme ical bu supe icially impe ec sequences
migh lead hem o pa sing hei s uc u e ins ead. Thus, in he p esen s udy we in es iga ed
— o he i s ime—whe he young in an s a 7 mon hs o age de ec symme ical s uc u es
in elabo a e isual sequences, speci ically, in mul i-colou ed mosaic-like abs ac isual pa -
e ns, o de e mine whe he in an s’ abili y o de ec symme y is main ained o dis up ed in
he absence o pe ec su ace symme y. This wo k is he e o e explo a o y in na u e.
The mosaics consis ed o colou ul squa e iles om wo dis inc ypes o ca ego ies, based
on bo h he shape o hei in e nal elemen s and hei colou combina ion (Fig 1). These iles
we e a anged in o ho izon al sequences, ei he in asymme ical o de o in e ical symme y.
We p esen ed 7-mon h-old in an s wi h mul iple ins ances o s uc u ally symme ical and
asymme ical mosaics and measu ed hei spon aneous looking imes a he wo ypes o pa -
e ns. We chose o examine 7-mon h-olds, as a his age in an s a e sensi i e o con as s in
shape and colou and, impo an ly, i is he younges age a which in an s a e known o g oup
s imuli in o la ge pe cep ual uni s on he basis o o m simila i y [28]. In e es ingly, adul s’
pe cep ion o symme y is acili a ed in pa e ns con aining elemen s g ouped in o clus e s
[10,29].
To de e mine he obus ness o symme y de ec ion in ea ly in ancy, we manipula ed wo
aspec s o ou s imuli, namely he leng h o he mosaic-like sequences and hei le el o
abs ac ness, de e mined by he elemen s o e which he s uc u al symme y holds (Fig 1).
The mosaics we e buil om he conca ena ion o ei he 3 o 5 iles, and we e s uc u ally sym-
me ical a ei he he oken- o he ca ego y-le el, i.e. symme ical a he le el o he speci ic
iles used ( epea ing a ile), o a he mo e abs ac le el o ile ca ego y, wi h no epe i ion o
speci ic iles. Symme ical mosaics had an unde lying s uc u e o de ed in e ical mi o sym-
me y: 5- ile sequences had an ABABA s uc u e, 3- ile sequences an ABA s uc u e, whe e A
and B ep esen he wo ca ego ies o iles. In bo h 3- and 5- ile sequences, he cen al A ele-
men aligns wi h he e ical axis o he sequence, and he le and igh hal es o he s uc u e
a e mi o p ojec ions. In summa y, s imuli wi h oken-le el symme ies epea ed speci ic
iles, while ca ego y-le el symme ies in ol ed only ile- ype, wi h no epe i ion (see Fig 1).
Impo an ly, none o he mosaics had pe ec su ace mi o symme y. All p io s udies exam-
ined in an s’ pe cep ion o su ace symme y. The p esen s udy is hence he i s one o exam-
ine whe he in an s de ec s uc u al symme y in he absence o pe ec su ace symme y.
We p edic ha di e ences in looking imes o he s uc u ally symme ical and asymme i-
cal pa e ns will e eal in an s’ isual p e e ences. P edic ing he di ec ion o in an s’ esponses
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 3 / 14
is, howe e , no s aigh o wa d. In an s migh look longe o he symme ical pa e ns i hey
p e e s uc u al symme y as a isual p ope y in complex isual sequences. Indeed, p e ious
s udies ha e shown ha in an s, child en, and adul s look longe a symme ical han asym-
me ical pa e ns when hese a e p esen ed side-by-side (in an s a 4 mon hs: [25], bu see also
[26]; in an s a 12 mon hs: [26]; 3- o 6-yea -old child en and adul s: [30]. Howe e , when p e-
sen ed wi h one display a a ime, 4-mon h-old in an s habi ua ed as e o symme ical pa -
e ns and hence looked longe du ing he p esen a ion o asymme ical designs [26]. Since ou
s imuli we e simila ly p esen ed one display a a ime, we p edic ed longe looking imes o
s uc u ally asymme ical mosaics.
Finding ha in an s de ec s uc u al symme y in ou complex s imuli, which a e he mos
ecological s imuli p esen ed in such expe imen s o da e o young in an s, will u he ou
knowledge o hei pa e n pa sing abili ies. I will also o e come wo impo an limi a ions o
p e ious s udies. Fi s , hese ea ly wo ks had e y educed simple sizes. Fu he mo e, hei
conclusions we e d awn on he basis o pai wise compa isons (i is no epo ed whe he hey
we e co ec ed o mul iple compa isons), a he han on Analyses o Va iance ac oss g oups
o measu emen s, wi h one excep ion [25]. Ou s udy will p o ide a me hodologically and s a-
is ically mo e solid basis o explo ing young in an s’ pe cep ion o symme y in complex
pa e ns.
Me hodology
Pa icipan s
Nine y-eigh 7-mon h-old in an s pa icipa ed in he expe imen (51 gi ls; mean age: 7;02; SD:
13 days; age ange: 6;14–8;04). All in an s we e bo n ull- e m and we e being aised a ound
Fig 1. S imuli. The uppe panel depic s a sample o he 18 iles used o c ea e he mosaics, 9 pe ca ego y (A o B). The
middle panel depic s a sample o he mosaic-like sequences wi h oken-le el s uc u al symme ies, and he lowe panel
o he mosaics wi h ca ego y-le el s uc u al symme ies.
h ps://doi.o g/10.1371/jou nal.pone.0266938.g001
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 4 / 14
he Pa is a ea in F ance. Pa icipan s we e andomly so ed in o ou g oups ha di e ed only
in he s imuli hey saw du ing he s udy, as desc ibed in he nex sec ion. Thus, 25 o he
in an s we e included in G oup 1 (14 gi ls; mean age: 7;02; SD: 14 days; age ange: 6;15–8;04),
ano he 24 in an s pa icipa ed in G oup 2 (11 gi ls; mean age: 7;01; SD: 12 days; age ange:
6;15–8;00), 24 in an s ook pa in G oup 3 (16 gi ls; mean age: 6;29; SD: 13 days; age ange:
6;14–8;02), and he emaining 25 in an s ook pa in G oup 4 (10 gi ls; mean age: 7;05; SD: 13
days; age ange: 6;15–7;27). Da a om 17 addi ional in an s we e no included due o ussiness
o c ying (6 in an s in G oup 1, 8 in G oup 2, 1 in G oup 3, and 2 in G oup 4), and 1 due o
equipmen ailu e (G oup 4). All pa en s ga e in o med consen be o e hei in an ’s
pa icipa ion.
The p e ious s udies examining young in an s’ su ace symme y de ec ion abili ies do no
epo e ec sizes o o he s a is ics ha would allow us o un a powe analysis. Sample size
was hence decided on he basis o in an a ailabili y and a ecen s udy by Oakes [31], examin-
ing he ade-o be ween sample size and s a is ical powe in in an looking- ime esea ch.
This s udy showed ha sample sizes below 24 in an s can be unde powe ed. We hus aimed o
include a leas 24 in an s in each g oup.
S imuli. S imuli we e 18 squa e-shaped, mul i-colou ed iles o iden ical size, akin o
hose used in [32]. Tiles we e spli in o wo ca ego ies: iles in ca ego y A con ained a ounded
shape and we e colou ed black, b own and blue, while iles in ca ego y B con ained angula
shapes and we e colou ed ed/o ange/pink and g een (see Fig 1). We combined he A and B
iles in o mosaic-like sequences o wo ypes: s uc u ally symme ical and asymme ical. The
symme ical sequences ollowed a simple ule o s ic al e na ion and had wo possible
leng hs: 3 iles (i.e. ABA) o 5 iles (i.e. ABABA). All esul ing sequences had an unde lying
bila e ally symme ic s uc u e along he e ical axis. Sequences could be s uc u ally symme -
ic ei he a (1) he oken le el, i.e. each sequence con ained a single A and B oken: A
a
B
1
A
a
(G oup 1) o A
a
B
1
A
a
B
1
A
a
(G oup 2), o (2) a he ca ego y le el, i.e. sequences con ained di -
e en okens o he same ca ego y: A
a
B
1
A
b
(G oup 3) o A
a
B
1
A
b
B
2
A
c
(G oup 4; see Fig 1).
While concep ually hese s imuli ep esen bila e al mi o symme y along a cen al e ical
axis, isually hei su ace symme y was no pe ec . In he case o mosaics wi h ca ego y-le el
symme y his is a necessa y esul o hei c ea ion (because iles in co esponding posi ions
a e di e en ). In mosaics wi h oken-le el s uc u al symme y, he s imuli a e mo e nea ly
symme ical, bu s ill iola e isual mi o symme y a a ine-g ained le el o de ail, by copy-
ing a he han mi o ing he epea ed A and B iles (see Fig 1). P io s udies show ha in an s
de ec su ace mi o symme y [24–27]. The p esen s udy examines whe he in an s also
de ec s uc u al mi o symme y in he absence o pe ec su ace symme y. I in an s di e -
en ia e be ween s uc u ally symme ic and asymme ic sequences, his would indica e ha
hey a e p ocessing he unde lying concep ual symme y o he symme ic pa e ns. Al e na-
i ely, hey migh be applying a global le el o pa sing ha dis ega ds hese small-scale de ia-
ions om pe ec isual symme y.
In o de o c ea e asymme ic a ian s o hese mosaics, we swi ched he o de o a pai o
adjacen iles wi hin he sequences, ensu ing ha all possible o de s occu ed wi h he same
equency. The 8 ABA sequences we e eo de ed in o 4 BAA and 4 AAB asymme ic
sequences ( he wo unde lined iles a e swapped). The 8 ABABA sequences esul ed in 2
BAABA, 2 AABBA, 2 ABBAA and 2 ABAAB sequences. The exhaus i e combina ion o he
wo manipula ed pa ame e s, sequence leng h (3 o 5) and le el o symme y (ca ego y o
oken), gene a ed 4 inal se s o mosaics, each con aining a o al o 8 symme ical and 8 asym-
me ical sequences. Each o hese 4 se s was es ed in a di e en g oup o in an s (G oups 1–4,
see de ails below).
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 5 / 14
P ocedu e. The s udy ook place a he Babylab o he In eg a i e Neu oscience and Cog-
ni ion Cen e (CNRS & Uni e si e
´Pa is Ci e
´) in Pa is, F ance, and was app o ed by he
CERES e hics boa d (Uni e si e
´Pa is Ci e
´). In an s we e sea ed on a pa en ’s lap in a sound-
a enua ed oom wi h dim ligh s. A ideo came a placed abo e he sc een eco ded he session.
Ca egi e s wo e opaque glasses, p e en ing hem om iewing he s imuli, in o de o a oid
po en ial pa en al in luence on he in an s. An expe imen e —placed ou side he es ing boo h
and blind o he s imuli—moni o ed in an s’ looking beha iou and con olled s imulus p e-
sen a ion. S imuli we e displayed using Habi X.10 so wa e [33], on a 23” LCD moni o . Ho i-
zon ally, he mosaics co e ed he o al wid h o he sc een. Consequen ly, 3- ile mosaics
occupied g ea e e ical space han 5- ile mosaics on sc een, as bo h ypes o sequences we e
gene a ed using he same squa e iles.
Tes consis ed o 16 ials: 8 con ained s uc u ally symme ical sequences and he emain-
ing 8 con ained s uc u ally asymme ical sequences. Al hough his numbe o ials is a he
high o s udies wi h such young in an popula ions, we easoned ha manipula ing he mosa-
ics’ leng h and le el o s uc u al symme y could esul in di e ing ajec o ies o symme y
de ec ion. We none heless designed he i s 8 ials o con ain 4 symme ical and 4 asymme i-
cal sequences o allow o he assessmen o looking p e e ences o e ewe ials in his ini ial
pe iod. O de o p esen a ion was addi ionally pseudo andomized so ha no mo e han wo
ials o he same ype occu ed consecu i ely. T ial o de also a ied ac oss babies.
In an s in G oup 1 saw 3- iled sequences wi h oken-le el s uc u al symme y. In an s in
G oup 2 saw 5- iled sequences wi h oken-le el s uc u al symme y. G oup 3 was p esen ed
wi h 3- iled sequences wi h ca ego y-le el s uc u al symme y. Finally, G oup 4 saw 5- iled
sequences wi h ca ego y-le el s uc u al symme ies (see Fig 1).
The s udy s a ed wi h a p e- es ial—a looming ball ha changed colou accompanied by
a woman’s oice saying “coucou”—in o de o a ac he in an ’s a en ion. Fu he mo e, each
ial began wi h ano he a en ion-ge e , i.e. a ideo showing lashing balls accompanied by a
bell sound (see Fig 2). Once he in an looked a he sc een, his ideo disappea ed and was
eplaced by one o he mosaics, which was p esen ed in silence. In an s saw comple e s imuli—
i.e. all iles wi hin a gi en sequence we e p esen ed simul aneously—which emained on
sc een o maximally 30 seconds o un il he in an looked away o mo e han 2 seconds.
A e his, a new ial began. A pos - es ial—iden ical o he p e- es ial— ollowed es .
Da a analysis. To es whe he in an s disc imina ed he s uc u ally symme ical and
asymme ical mosaics, we eco ded hei spon aneous a en ion o he sc een du ing he 16
es ials and coded hei looking beha iou o -line. Two esea ch assis an s, blind o he con-
di ions, coded hal o he in an s each. In addi ion, bo h assis an s coded 8 andomly chosen
in an s, o measu e he eliabili y o hei coding. Code s achie ed a high le el o ag eemen (
= .96; p <.001). As is cus oma y in s udies analysing in an looking beha iou , we excluded
om analysis all ials wi h e y sho (<1sec) looking imes [34]. A e applying his c i e ion,
only in an s ha had a minimum o h ee ials pe condi ion—s uc u ally symme ical and
asymme ical mosaics—we e e ained o analysis. Implemen a ion o hese c i e ia did no
esul in he exclusion o any babies om analysis. O he o al o 1568 ials (98 in an s x 16 i-
als each), 1530 en e ed analysis. The emaining 38 ials (2.42%) we e excluded due o: (1) ha -
ing looking imes sho e han 1sec (30 ials), (2) expe imen e e o du ing online coding (2
ials), (3) and pa en al in e e ence (6 ials). The sample o 98 in an s had a mean numbe o
7.78 symme ical ials ou o 8 ( ange 5 o 8) and 7.84 in asymme ical ials ( ange 5 o 8).
The ull se o da a is a ailable in he S1 File.
We analysed in an s’ a e age looking imes o he s uc u ally symme ical and asymme i-
cal ials and s a is ically e alua ed he e ec s o s uc u al symme y, sequence leng h, le el o
symme y and a iabili y in looking imes. In o de o de ec po en ially di e en ajec o ies
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 6 / 14
ac oss g oups, we spli he es phase in o wo hal es, as is o en done in he li e a u e (e.g. [35–
38]).
Resul s
We a e aged he in an s’ looking imes ac oss all ials o he same condi ion—i.e. s uc u ally
symme ic o asymme ic—du ing he i s and second hal es o he s udy (see Fig 3 and
Table 1), and ca ied ou a epea ed-measu es ANOVA wi h looking ime as he dependen
a iable, S imulus Type (s uc u ally symme ic o asymme ic) and Block ( i s 8 ials s. las
8 ials) as wi hin-subjec a iables, as well as Sequence Leng h (3 o 5 iles) and Symme y
Type ( oken- o ca ego y-le el) as be ween-subjec a iables. The ANOVA yielded a signi ican
main e ec o Block (F(1, 94) = 55.101, p <.001,
p2
= .370, 95% CI o he di e ence [2.53,
4.38]), due o longe o e all looking imes du ing he i s han in he second hal o he s udy.
In addi ion, he ANOVA e ealed a signi ican in e ac ion be ween Block and S imulus Type
(F(1,94) = 4.801, p = .031,
p2
= .049). No o he e ec s o in e ac ions eached signi icance (all
ps�.085, he esul s o he ANOVA a e de ailed in he S2 File).
In o de o explo e he signi ican in e ac ion be ween block and s imulus ype, we ca ied
ou sepa a e ANOVAs on he i s and las 8 ials o he s udy (see Fig 3 and Table 1), wi h
S imulus Type, Sequence Leng h and Symme y Type as a iables. The ANOVA on he i s 8
ials o he s udy e ealed a main e ec o S imulus Type (F(1, 94) = 5.498, p = .021,
p2
= .055,
95% CI o he di e ence [0.15, 1.76]), due o longe o e all looking imes o s uc u ally asym-
me ical han symme ical mosaics. The e we e no u he e ec s o in e ac ions (all ps�
.255). In u n, he ANOVA on he las 8 ials o he s udy e ealed no signi ican e ec s o
in e ac ions (all ps�.121). The esul s o he wo ANOVAs a e epo ed in he S2 File.
Ou esul s indica e ha 7-mon h-old in an s disc imina ed s uc u ally symme ical om
asymme ical mosaics, al hough his e ec disappea ed as in an s’ a en ion—as measu ed by
looking ime—declined du ing he cou se o he expe imen . In an s hus appea o ha e pe -
cei ed s imulus s uc u al symme y, and hei looking beha iou was no modula ed by wo
addi ional dimensions o a iabili y p esen in he mosaics, i.e. s imulus leng h and symme y
ype. No e ha , ac oss g oups, mosaics also di e ed in a hi d sou ce o a iabili y as a neces-
sa y consequence o he i s wo, i.e. he numbe o unique iles p esen in each mosaic. Mosa-
ics wi h oken-le el s uc u al symme ies—A
a
B
1
A
a
and A
a
B
1
A
a
B
1
A
a
—consis ed o wo
Fig 2. P ocedu e o he es . The s udy s a ed wi h a p e- es ial in o de o engage in an s’ a en ion. In an s hen saw a o al o 8
s uc u ally symme ical and 8 s uc u ally asymme ical sequences, which we e in e mixed and p eceded by an a en ion ge e . A
pos - es ial iden ical o he p e- es ial ended he s udy.
h ps://doi.o g/10.1371/jou nal.pone.0266938.g002
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 7 / 14
unique iles, one pe ca ego y (i.e. A
a
and B
1
). Meanwhile, mosaics wi h ca ego y-le el s uc-
u al symme ies—A
a
B
1
A
b
and A
a
B
1
A
b
B
2
A
c
—con ained ei he 3 (i.e. A
a
, A
b
, B
1
) o 5 unique
iles (i.e. A
a
, A
b
, A
c
, B
1
, B
2
). In o de o de e mine whe he his speci ic sou ce o a iabili y
impac ed in an s’ disc imina ion o he s uc u ally symme ical and asymme ical mosaics,
we an an addi ional epea ed-measu es ANOVA wi h Numbe o Unique Tiles (2, 3 o 5) as a
be ween-subjec s ac o , S imulus Type (symme ic o asymme ic) and Block ( i s 8 ials s.
las 8 ials) as wi hin-subjec s ac o s, and looking imes as he dependen a iable. Once
again, a main e ec o Block ob ained (F(1, 95) = 49.114, p <.001,
p2
= .341, 95% CI o he di -
e ence [2.46, 4.40]), as well as an in e ac ion be ween Block and S imulus Type (F(1, 95) =
6.024, p = .016,
p2
= .060), bu no e ec o Numbe o Tiles (p = .306) o in e ac ion (all ps�
.159, he esul s o he ANOVA a e epo ed in he S2 File).
Discussion
We in es iga ed whe he young in an s pe cei e s uc u al symme y in mul i- ea u ed
abs ac isual pa e ns, p esen ing 7-mon h-old in an s wi h images o colou ul mosaics buil
om squa e iles om wo ca ego ies—A and B—based on hei colou and in e nal shape.
Fig 3. Looking ime esul s. The uppe ba plo shows in an s’ mean looking imes o he mosaics: du ing he 16 ials (le ), in he i s 8 ials
(cen e), and in he las 8 ials ( igh ). Looking imes in all 4 condi ions (i.e. o 3- and 5- iled mosaics, wi h oken- and ca ego y-le el s uc u al
symme y) a e collapsed. The y-axis displays he in an s’ looking imes in seconds. Da k blue ba s depic mean looking imes o s uc u ally
symme ical mosaics, while ligh aquama ine ba s display mean looking imes o s uc u ally asymme ical mosaics. E o ba s ep esen he s anda d
e o o he mean, and s a is ically signi ican compa isons a e ma ked wi h an as e isk. The lowe box-and-whiske plo depic s he dis ibu ion o
in an s’ mean looking imes pe condi ion du ing he i s 8 ials.
h ps://doi.o g/10.1371/jou nal.pone.0266938.g003
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 8 / 14
Tiles we e a anged in o mosaic-like sequences wi h a s uc u ally symme ical (e.g. ABA,
ABABA) o asymme ical s uc u e (e.g. AAB, ABAAB). We measu ed in an s’ spon aneous
looking beha iou o bo h ypes o mosaics, manipula ing wo p ope ies o he ile-sequences,
namely hei leng h and abs ac ness o symme y. Mosaics could consis o sequences o ei he
3 o 5 iles (e.g. ABA o ABABA), and be s uc u ally symme ical a ei he he oken le el (i.e.
he indi idual okens we e iden ical) o a he ca ego y le el (i.e. only he ypes o okens we e
he same). While he symme ical sequences had mi o symme y a he s uc u al le el, nei-
he oken no ca ego y le el mosaics had pe ec su ace symme y.
P e ious li e a u e examining in an s’ pe cep ion o symme y used simple designs han
ou mul i- ea u ed mosaics, such as a angemen s o a ew do s, o a simple shape o pa e n,
all o hem mono-ch oma ic [24–27] and wi h pe ec su ace symme y, making ou s udy
wi h i s complex and colou ul s imuli explo a o y in na u e, and he i s one o examine
whe he in an s de ec s uc u al symme y in he absence o pe ec su ace symme y. We
op ed o p esen in an s wi h 16 ials, easoning ha manipula ing he leng h o he mosaics
and/o hei le el o s uc u al symme y could esul in di e ing ajec o ies o symme y
de ec ion. Analysis o he 16 ials unco e ed no signi ican di e ence in in an s’ looking imes
o he s uc u ally symme ical and asymme ical mosaics. Howe e , in an s’ a en ion decayed
signi ican ly as he s udy p og essed, which sugges s ha he leng h o he s udy was excessi e
o such young in an s. In addi ion, we obse ed a signi ican in e ac ion be ween in an s’
Table 1. Looking ime esul s.
ALL 16 TRIALS
Symme ical mosaics Asymme ical mosaics
mean SE mean SE
oken-le el symme y G oup 1: ABA 8.90 0.73 8.64 0.81
G oup 2: ABABA 7.83 0.75 8.68 0.83
ca ego y-le el symme y G oup 3: ABA 7.09 0.75 7.17 0.83
G oup 4: ABABA 7.50 0.73 8.34 0.81
mean o all g oups 7.83 0.37 8.21 0.41
FIRST 8 TRIALS
Symme ical mosaics Asymme ical mosaics
Mean SE mean SE
oken-le el symme y G oup 1: ABA 10.15 0.90 10.49 1.09
G oup 2: ABABA 9.83 0.92 10.63 1.11
ca ego y-le el symme y G oup 3: ABA 8.15 0.92 8.92 1.11
G oup 4: ABABA 8.94 0.90 10.85 1.09
mean o all g oups 9.27 0.46 10.22 0.55
LAST 8 TRIALS
Symme ical mosaics Asymme ical mosaics
Mean SE mean SE
oken-le el symme y G oup 1: ABA 7.64 0.87 6.79 0.87
G oup 2: ABABA 5.82 0.89 6.73 0.88
ca ego y-le el symme y G oup 3: ABA 6.02 0.89 5.42 0.88
G oup 4: ABABA 6.07 0.87 5.83 0.87
mean o all g oups 6.39 0.44 6.19 0.44
Mean looking imes and s anda d e o o he mean (SE) in seconds, o s uc u ally symme ical s. asymme ical mosaics, in he ou g oups o in an s. The uppe panel
displays mean looking imes including all 16 ials, he cen al panel con ains mean looking imes o he i s 8 ials only, and he lowe panel displays mean looking
imes du ing he las 8 ials.
h ps://doi.o g/10.1371/jou nal.pone.0266938. 001
PLOS ONE
Se en-mon h-old in an s de ec symme ical s uc u es in mul i- ea u ed abs ac isual pa e ns
PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0266938 May 11, 2022 9 / 14
