Exploring eye-movement changes as digital biomarkers and endophenotypes in subclinical eating disorders: an eye tracking study

Na as‑Leóne al. BMC Psychia y (2025) 25:133
h ps://doi.o g/10.1186/s12888‑025‑06583‑z
RESEARCH Open Access
© The Au ho (s) 2025. Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which
pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e app op ia e c edi o he
o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons licence, and indica e i changes we e made. The images o
o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line
o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed by s a u o y
egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his
licence, isi h p://c ea i ecommons.o g/licenses/by/4.0/.
BMC Psychia y
Explo ing eye‑mo emen changes asdigi al
bioma ke s andendopheno ypes insubclinical
ea ing diso de s: aneye acking s udy
Se gio Na as‑León1,2 , Milag osa Sánchez‑Ma ín1 , Ana Tajadu a‑Jiménez3,4 , Lize De Cos e 3,5 ,
Me cedes Bo da‑Mas6 and Luis Mo ales1*
Abs ac
Objec i e P e ious esea ch has indica ed ha pa ien s wi h Ano exia Ne osa (AN) exhibi speci ic eye mo emen
changes, iden i ied h ough eye acking senso echnology. These changes ha e been p oposed as po en ial digi al
bioma ke s and endopheno ypes o ea ly diagnosis and p e en i e clinical in e en ions. This s udy aims o explo e
whe he hese eye mo emen changes a e also p esen in indi iduals wi h subclinical ea ing diso de (ED) symp oma‑
ology compa ed o con ol pa icipan s.
Me hod The s udy ec ui ed pa icipan s using con enience sampling and employed he Ea ing Diso de Examina‑
ion Ques ionnai e o ini ial sc eening. The sample was subsequen ly di ided in o wo g oups: indi iduals exhibi ing
subclinical ED symp oma ology and con ol pa icipan s. Bo h g oups pe o med a ious asks, including a ixa ion
ask, p osaccade/an isaccade ask, and memo y‑guided ask. Alongside hese asks, anxie y and p emo bid in el‑
ligence we e measu ed as po en ial con ounding a iables. The da a we e analyzed h ough means compa ison
and explo a o y Pea son’s co ela ions.
Resul s No signi ican di e ences we e ound be ween he wo g oups in he h ee eye acking asks.
Discussion The indings sugges ha he obse ed changes in p e ious esea ch migh be mo e ela ed o he clini‑
cal s a e o he illness a he han a pu a i e ai . Implica ions o he applicabili y o eye mo emen changes as ea ly
bioma ke s and endopheno ypes o EDs in subclinical popula ions a e discussed. Fu he esea ch is needed o ali‑
da e hese indings and unde s and hei implica ions o p e en i e diagnos ics.
Regis a ion h ps:// jea d iso d. biome dcen al. com/ a ic les/ 10. 1186/ s40337‑ 022‑ 00573‑2
Keywo ds An isaccade, Ea ing diso de s, Inhibi o y con ol, Memo y‑guided saccade, P osaccade, Saccades,
Subclinical popula ion, Squa e wa e je ks, Visual memo y, P e en ion
*Co espondence:
Luis Mo ales
[email p o ec ed]
Full lis o au ho in o ma ion is a ailable a he end o he a icle
Page 2 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
Backg ound
Ea ing diso de s (EDs) can be di icul o de ec in ou-
ine psychia ic ca e [1–3]. To sol e his p oblem,
g owing e idence is ocusing on he iden i ica ion o bio-
ma ke s and endopheno ypes associa ed wi h EDs [4, 5].
Bioma ke s a e objec i ely measu ed indica o s o bio-
logical p ocesses, whe eas endopheno ypes a e quan i a-
i e ai s ha lie in he causal pa hway om geno ype o
pheno ype [6]. While endopheno ypes ha e a causa i e
ole, bioma ke s a e me ely isk indica o s [6]. Despi e
hese di e ences, bo h can enhance popula ion-based
sc eening o iden i ying indi iduals a isk o psycho-
pa hological diso de s and hose who could bene i om
p e en i e ea men s [6].
In his ega d, P eselle e al. [7] conduc ed a sys em-
a ic e iew ha iden i ied a o al o 141 s udies u ilizing
a a ie y o senso s o measu e aspec s ela ed o EDs.
The e iew ound ha senso s can objec i ely assess
some EDs ele an beha io s such as physical ac i i y,
ood in ake, and sleep pa e ns. Because EDs a e cha -
ac e ized by a mul i ace ed symp oma ology, his line o
esea ch holds p omise as a complemen a y ool ha can
enhance ou unde s anding o EDs and aid in iden i y-
ing indi iduals a isk who may bene i om p e en a i e
in e en ions, con ibu ing o mo e a o able he apeu ic
ou comes [7–11].
In his con ex , s udies using eye acking (i.e., senso
echnology ha measu es eye posi ions and mo emen s)
ha e highligh ed a possible a en ional bias linked o dis-
o de - ela ed s imuli, such as ood o bodies, dis inc i e
beha io al bioma ke s, sugges ing an al e ed isual a en-
ion in indi iduals wi h EDs [6]. Howe e , ew s udies
ha e speci ically add essed he ole o speci ic eye mo e-
men s pa e ns o diso de -un ela ed s imuli [12]. In his
line, Phillipou e al. [13] ound ha pa ien s wi h Ano-
exia Ne osa (AN; an ED cha ac e ized by es ic ed
ood in ake and an in ense ea o gaining weigh )
showed poo pe o mance on ocula ixa ions asks, i.e.,
an impai ed abili y o main ain ixa ion on a single do
compa ed wi h con ol pa icipan s. Speci ically, hey
showed an inc eased a e o saccadic in usions named
squa e wa e je ks (SWJs; hese a e ho izon al, in olun-
a y, saccadic in usions ha in e up ixa ion). Recen ly,
Phillipou e  al. [14] iden i ied he s a e independence
and he i abili y o his possible bioma ke . Acco ding o
he esul s, hey ound ha pa ien s wi h AN, pa ien s’
weigh - es o ed om he illness, and sis e s o people
wi h AN, made signi ican ly mo e SWJs han heal hy con-
ols. Fu he mo e, he combina ion o SWJ a e and anx-
ie y showed high accu acy le els o disc imina e be ween
he di e en g oups es ed (≥ 70%). Ano he s udy using
p osaccade/an isaccade and memo y-guided saccade
asks showed he exis ence o speci ic eye mo emen
anomalies in pa ien s wi h AN [15]. In he p osaccade
ask pa icipan s a e ins uc ed o ix on a cen al do .
Then, hey ha e o di ec hei gaze owa d a a ge do
appea ing a he pe iphe y as quickly and as accu a ely as
possible. This ask is s imulus-d i en as i equi es o pe -
o m a saccade (i.e., a ballis ic mo emen o he eyes ha
shi s he cen e o gaze) o an onse pe iphe al s imulus,
a p ocess which is di icul o inhibi . On he con a y, in
he an isaccade ask pa icipan s a e ins uc ed o look in
he opposi e di ec ion o he pe iphe ical do . This ask
is goal-d i en since i equi es oli ional p ocessing, a
olun a y saccade o he opposi e di ec ion o he s imu-
lus. The memo y-guided saccade ask equi es a saccade
owa ds a emembe ed onse pe iphe al s imulus a e a
b ie delay. Compa ed o con ol pa icipan s, pa ien s
wi h AN ended o show sho e saccade la encies in he
p osaccade/an isaccade ask and mo e inhibi o y e o s
in he memo y-guided saccade ask [15]. In a mo e ecen
s udy, Phillipou e al. [16] epo ed ha people wi h AN
made mo e inhibi o y e o s han he weigh - es o ed
AN g oup, sis e s o people wi h AN, and heal hy con-
ols. The indings demons a e a po en ial s a e-depend-
en measu e associa ed wi h he ac i e clinical s a us o
he illness [16].
Ne e heless, he e is a call o ocus no only on AN
sample, bu o also include di e se samples such as o he
ED sub ypes [5]. Acco ding o his ansdiagnos ic pe -
spec i e, i would be possible o ind po en ial bioma k-
e s o endopheno ypes o iden i y "pa icula symp oms
ac oss he e ogeneous illness p o iles” [5]. Gi en he
cu en e idence, i is no ye known whe he hese eye
mo emen pa e ns s a o mani es also in he e ogene-
ous samples wi h subclinical ED symp oma ology. This
line o esea ch is p omising as i sugges s he possi-
ble use o emo e and po able eye acke s as an addi-
ional ool o sc een o subclinical ED symp oma ology
in p ima y ca e, complemen ing s anda d assessmen
p ac ices. This could con ibu e o enhancing p ognos-
ic and p e en i e ea men s, an a ea ha wa an s u -
he explo a ion in he ield [5, 10]. Fu he mo e, s udies
using subclinical ED samples could help o shed ligh on
whe he hese changes may e lec he condi ion i sel , a
ai , o a s a e o he illness du ing he ac i e phase.
S udy objec i es andhypo hesis
The p ima y aim was o in es iga e whe he young adul
women wi h subclinical ED symp oma ology and con-
ol pa icipan s di e in hei pe o mance on se e al
eye mo emen asks. We hypo hesize ha pa icipan s
wi h subclinical ED symp oma ology compa ed o con-
ols would be p one o show sho e saccade la encies
and mo e inhibi o y e o s in a p osaccade/an isaccade
ask and in a memo y-guided saccade ask, espec i ely,
Page 3 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
and an inc eased a e o SWJs in a ixa ion ask. As a sec-
onda y aim, we es ed he po en ial ole o SWJ a e and
anxie y o disc imina e he di e en g oups es ed in case
he e we e signi ican di e ences (as in [14]).
Me hods/design
The s udy was conduc ed in acco dance wi h he Decla a-
ion o Helsinki and a e ob aining local E hics Commi -
ee app o al (Re : PID2019-105579RB). The expe imen al
p o ocol o his s udy can be ound in [6].
Pa icipan s andp ocedu e
A young adul emale sample was chosen due o he
highe p e alence o EDs in his popula ion [17–19].
Thus, he sample comp ised young adul women in he
18–25 age ange. The sample p ocedu e was h ough
public ad e isemen s and social media pos s (con eni-
ence sampling). Inclusion c i e ia we e: (a) no mal (o
co ec ed) isual acui y; (b) sex a bi h: women; (c) cu -
en BMI in he no mal ange acco ding o Wo ld Heal h
O ganiza ion (WHO) (be ween 18.5– 24.9). Exclusion
c i e ia: (a) sel - epo ed li e ime his o y o signi ican
b ain inju y, neu ological condi ion, ocula and/o is-
ual pa hology; (b) sel - epo ed li e ime his o y o an
ED o o he men al illness; (c) sel - epo ed cu en use
o psycho opic d ugs (e.g., an idep essan s) o in ake
o ec ea ional syn he ic o na u al d ugs; (d) incom-
ple e da a collec ion o eye- acke calib a ion ailu e;
(e) inabili y o unde s and Spanish; ( ) ou o ange age.
The Spanish Ea ing Diso de Examina ion Ques ionnai e
(S-EDE-Q) [20] was adminis e ed since i is conside ed
he gold s anda d o assessmen o ED pa hology [21].
The EDE-Q has shown good p edic i e as well as concu -
en alidi y [20, 22]. Fu he , i has been shown o ha e
good psychome ic p ope ies in young adul s in Spain
[20]. This ool se ed as a p e-sc eening ool, allowing us
o spli he sample in o wo g oups o pa icipan s wi h
subclinical ED symp oma ology and con ol pa ici-
pan s acco ding o hei EDE-Q sco e, as de ined by he
s udy. Following he sugges ed cu -o o 4 o Fai bu n
and Coope [23], which has been widely used in p e i-
ous esea ch wi h bo h clinical [24, 25] and non-clinical
samples [26–28], he esponses on bo h EDE-Q i ems 22
(Weigh o e alua ion: “Has you weigh in luenced how
you eel abou you sel as a pe son?”) and 23 (Shape o e -
alua ion: “Has you shape in luenced how you eel abou
you sel as a pe son?”) we e used. Pa icipan s we e
classi ied wi h subclinical ED symp oma ology i hey
a ed he a o emen ioned EDE-Q i ems as 4 o highe
on a scale o 0 (no a all) o 6 (ma kedly). On he con-
a y, pa icipan s we e classi ied wi hou subclinical ED
symp oma ology (con ol g oup) i hey a ed hese i ems
below 4.1
Recen esea ch suppo s he use o weigh /shape
impo ance as a meaning ul cu o , pa icula ly ac oss
di e en ED diagnoses [29–31] and has been ex ensi ely
employed in p io esea ch [32]. This app oach aligns
wi h he ansdiagnos ic model o EDs [33], which posi s
ha EDs sha e a common co e psychopa hology in ol -
ing he o e alua ion o body shape and weigh [34, 35].
The p e-sc eening allowed us o o m wo g oups o simi-
la size: once he desi ed pa icipan sample was eached
o one o he g oups, only pa icipan s alling in o he
o he g oup de ined o he s udy we e in i ed o ake
pa .
A he beginning o he expe imen w i en in o med
consen was ob ained om all pa icipan s. Subsequen ly,
o iden i y possible con ounds, we measu ed a iables
ha may ha e a ec ed pe o mance in he eye mo emen
asks (p emo bid in elligence, i.e., be o e he onse o he
diso de , and anxie y) [15]. Only hose wi h signi ican
ela ionships be ween g oups will be deemed po en ial
con ounde s and included in he analysis.
Nex , a ba e y o eye mo emen asks p esen ed in he
same o de was adminis e ed (see he Measu es sec ion
o mo e de ails). A e comple ing he s udy, a deb ie ing
session was ca ied ou . The ull p ocedu e ook app oxi-
ma ely 90min.
Measu es
Psychome ic measu es
Ea ing diso de symp oma ology: Spanish Ve sion
o he Ea ing Diso de Examina ion Ques ionnai e
(S‑EDE‑Q) [20] Composed by 28 7-poin Like - ype
esponse i ems anging om 0 (no a all) o 6 poin s
(ma kedly). Fou subscales a e measu ed, including die-
a y es ain , shape conce ns, weigh conce ns, and ea -
ing conce ns. The global sco e anges om 0 o 6 poin s.
Highe sco es on he EDE-Q indica e highe le els o
ED symp oma ology. The subscales demons a ed high
eliabili y: he Res ain subscale had an alpha o 0.829,
Ea ing Conce n was 0.827, Shape Conce n was 0.943,
and Weigh Conce n was 0.867. The global sco e o he
S-EDE-Q exhibi ed excep ional eliabili y wi h an alpha
o 0.961.
1 I should be no ed ha he o iginal c i e ion o de ec ing EDs in he p o o-
col published was based on a global S-EDE-Q sco e o ≥ 2.8, as sugges ed by
he s udy o Mond e al. [21]. Howe e , employing a cu o sco e o 4 o highe
o he impo ance o shape o weigh aligns mo e closely wi h he ansdiag-
nos ic non-clinical na u e o ou sample. This app oach is pa icula ly pe i-
nen gi en ha he o iginal alida ion o he 2.8 cu o sco e was con ined o
popula ions diagnosed exclusi ely wi h AN and BN in p ima y ca e [22].
Page 4 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
P emo bid in elligence
The Wo d Accen ua ion Tes [36] Composed by 30
low equency Spanish wo ds whose accen s ha e been
emo ed. Pa icipan s mus demons a e hei knowledge
o he co ec accen ua ion o each wo d. The o al sco e
is he numbe o wo ds co ec ly ead ( om 0 o 30). The
es is adminis e ed indi idually and akes 2–3min. I has
demons a ed excellen psychome ic p ope ies in e ms
o eliabili y and alidi y in he Spanish popula ion (del
Pino e al., 2018).
Anxie y le els
S a e‑T ai Anxie y In en o y (STAI) [37] The wo
o ms o anxie y (s a e and ai ) a e sepa a ed in he
in en o y, and bo h ha e hei own 20 sepa a e ques ions.
The ques ionnai e is he e o e composed o 40 i ems
anging om 0 (almos ne e ) o 4 (almos always). The
sco e anges om 0 o 80 poin s. Highe sco es on he
STAI indica e highe le els o anxie y. Fo bo h o ms,
highe sco es indica e highe le els o anxie y. The S a e-
T ai Anxie y In en o y (STAI) showed high eliabili y
o he s a e anxie y subscale (α = 0.937) bu a low eli-
abili y o he ai anxie y subscale (α = 0.483).
O he measu es
Age and highes le el o educa ion comple ed we e col-
lec ed. Body mass index (BMI) was also measu ed.
Da a acquisi ion
Pa icipan s we e in i ed o si in a b igh oom wi h
cons an ligh ing and empe a u e in on o a 24″
LED 144 Hz moni o (Asus VG248QE; 60-Hz e esh
a e) wi h esolu ion 1920 × 1080 pixels. An adjus able
chin es was used o minimize head mo emen s and o
ensu e a cons an dis ance be ween he pa icipan s’ eyes
and he sc een (90cm). Eye mo emen s we e eco ded
wi h he Eye-Link Po able Duo (SR Resea ch, On a io,
Canada). Bo h eyes we e eco ded a a sampling a e o
500Hz. The de ice uses a da k pupil- o-co nea e lec ion
me hod. A saccade eloci y h eshold o 30°/sec, an accel-
e a ion h eshold o 8000°/s2 and a mo ion h eshold o
0.15° was se . Be o e each ask, an au oma ic andomized
9-poin calib a ion was conduc ed on a black backg ound
sc een and d i co ec ion was pe o med h oughou
he ask when necessa y.
Eye mo emen asks
Fixa ion ask [13] The pa icipan is asked o look a a
cen ed ixa ion do (diame e o 0.5 deg ees) o 1min
(see Fig.1a). The ask comp ises h ee ials wi h a sho
es ing pe iod be ween hem.
Fig. 1 Eye‑ acking ask. a Fica ion ask, b P osaccade/ An isaccade, c Memo y‑ guided ask
Page 5 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
P osaccade/an isaccade ask [38] Each ial begins
wi h a cen ed ixa ion c oss ( andom pe iod 1000 –
3000ms) ollowed by a do (diame e o 0.5 deg ees) ha
appea s in a andom loca ion ei he 8º le o igh o he
ixa ion c oss (1000ms). In he p osaccade condi ion he
pa icipan is equi ed o di ec he gaze owa ds he do
as quickly as possible, while in he an isaccade condi ion
he pa icipan is equi ed o di ec he gaze o he oppo-
si e posi ion o he do (see Fig.1b). Fi s , pa icipan s
a e equi ed o comple e a p ac ice phase consis ing o
i e p osaccade ials and i e an isaccade ials. Once
p ac ice ials a e comple ed, a o al o 240 ials a e
p esen ed o ganized in i e blocks as ollows: 60 p osac-
cades; 40 an isaccades; 40 an isaccades; 40 an isaccades;
60 p osaccades (see Fig.1b).
Memo y‑guided saccade ask [15]
The pa icipan is asked o look a a cen ed ixa ion c oss
o a andom pe iod om 1000 o 3000ms. Nex , a 50ms
do (diame e o 0.5 deg ees) is p esen ed in a andom
loca ion in he pe iphe y (5–10° le o igh ) while he
cen al ixa ion emains o a andom pe iod in he in e -
al 1000–3000ms. The pa icipan is ins uc ed o keep
he eyes on his cen al ixa ion du ing he en i e delay
and o igge a saccade owa ds he memo ised loca ion
o he pe iphe al do as soon as he cen al ixa ion disap-
pea s. Fi s , pa icipan s a e equi ed o comple e a p ac-
ice phase consis ing o eigh ials. Once p ac ice ials
a e comple ed, a o al o 52 ials a e p esen ed, wi h an
equal numbe o a ge p esen a ions o each pe iphe al
loca ion (see Fig.1c).
Fo all he asks: c oss ixa ion colou , on ype and
size (RGB: 255,255,255; Fon : Times New Roman, 28).
Backg ound sc een colou (RGB: 51,51,51).
Da a analysis
Fo eye acking da a managemen , Excel 2021 ( e sion
18.0) was used, employing sc ip s in Excel Mac o lan-
guage. Likewise, o each eye mo emen analysis ask,
a speci ic sc ip was de eloped in he R p og amming
language [39], aking ad an age o i s ad anced analysis
capaci y and lexibili y. JASP ( e sion 0.16) was used o
all s a is ical analyses.
Sample size calcula ion
80% powe and a signi icance c i e ion o α = 0.05 we e
se since hese a e conside ed a con en ion o gene al
use speci ically in he psychology ield [40, 41]. Rega ding
he e ec size, based on p e ious li e a u e on he opic
[16], we choose a la ge e ec size (d = 0.80) [40]. Thus,
a p io i - es G Powe 3.1.9.2 (Dusseldo Uni e si y,
Ge many, h p:// www. gpowe . hhu. de/ en. h ml) analysis
yielded a minimum sample o 26 pa icipan s in each
g oup (N = 52, wo g oups: pa icipan s wi h subclinical
ED symp oma ology s. con ol pa icipan s).
S a is ical analysis
No mali y and ou lie s we e checked h ough g aphic
es s such as QQ plo s, his og ams, and box plo s. G oup
analyses was pe o med wi h S uden ’s - es pai ed
compa ison ( o no mal dis ibu ion) o Wilcoxon
es s ( o non-no mal dis ibu ion). We applied Benja-
mini–Hochbe g p ocedu e co ec ion o mul iple - es
compa isons.
In case o signi ican di e ences be ween s udy g oups,
we hen included con ounding a iables in he analysis
desc ibed abo e: p emo bid in elligence and anxie y was
conside ed a con ounde o all a iables based on p e i-
ous li e a u e (i.e., [13]).Ac oss g oups, explo a o y Pea -
son’s co ela ion analyses we e also conduc ed be ween
S-EDE-Q, STAI sco es, and BMI and he ou come meas-
u es associa ed wi h he eye acking da a (see Table1,
o a de ailed o e iew o he ou come measu es). Addi-
ionally, i signi ican di e ences we e ound be ween
STAI sco es and SWJ a e, consis en wi h p e ious li -
e a u e [13, 14], a disc iminan analysis would be pe -
o med on he en i e sample o elucida e o wha ex en
bo h measu es co ec ly classi y g oup membe ship (pa -
icipan s wi h subclinical ED symp oma ology s. con ol
pa icipan s).
Resul s
Fo he p osaccade/an isaccade ask, mos exclusions
we e due o ixa ion e o s (4.429%), which occu when
pa icipan s ail o main ain hei gaze on a p ede e -
mined poin . This was ollowed by exclusions due o
exceeding ampli ude h esholds (1.427%), whe e he
eye mo emen s exceed he de ined ange o mo ion o
he ask. The leas numbe o exclusions we e ela ed o
blinks (0.162%), which in ol e in olun a y o olun a y
eye closu es ha in e e e wi h da a collec ion. The o al
was 6.109% o ials excluded. Fo he memo y-guided
saccade ask, exclusions we e p ima ily due o exceed-
ing ampli ude h esholds (5.629%), ollowed by ixa ion
e o s (1.638%), and he leas numbe o exclusions we e
ela ed o blinks (0.524%). The o al was 7.762% o ials
excluded. A e ha , he pe cen age o ials excluded o
each eye mo emen ask was de e mined o each pa -
icipan , using his exclusion pe cen age as a c i e ion o
iden i ying ou lie cases wi h boxplo s. Based on his c i-
e ion, da a om se en pa icipan s we e excluded. On
he o he hand, no mul i a ia e ou lie s we e de ec ed
when applying he Mahalanobis dis ance o he psycho-
logical a iables collec ed in he ques ionnai es (i.e., ED
symp oma ology and anxie y). The o al sample size was

Page 6 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
Table 1 Ou come measu es associa ed wi h he eye‑ acking da a
P ac ice ials we e aimed o acquain he pa icipan wi h he expe imen al pa adigms and we e no s a is ically analyzed. In he saccade asks, saccades smalle han 2° we e no analyzed. The e o a e o each
pa icipan was calcula ed as he p opo ion o e oneous ials o all alid ials. The gain o he i s saccade was calcula ed as a a io o he i s saccade ampli ude di ided by he desi ed saccade ampli ude (e.g., 8° o
P osaccade ask). La ency o he i s saccade was de ined as he la ency om appea ance o he a ge o he s a o he saccade
Task Ou come
Fixa ion Ra e o SWJs. Th eshold o SWJ de ec ion included saccade pai s (an ini ial saccade ha mo es he o ea away om he c oss ixa ion, ollowed by a second saccade
in he opposi e di ec ion o e o ea e he ixa ion) wi hin 200 ms, wi h ampli udes be ween 0.1° and 5°
P osaccade Gain, la ency, and peak eloci y o co ec saccades E o s ‑An icipa ion a e: making a saccade du ing he ixa ion pe iod, p io o he p es‑
en a ion o he a ge do o wi hin 80 ms o i s p esen a ion
‑P osaccade e o : saccade in he opposi e di ec ion o he pe iphe al s imulus
An isaccade Gain, la ency, and peak eloci y o co ec saccades E o s ‑Co ec ed e o a e: looking a he a ge , hen looking a he co ec loca ion
a he
mi o image o he a ge
‑La ency o he co ec ion saccade: how long i ook o make a saccade in he co ‑
ec
di ec ion ollowing he inco ec saccade
‑Unco ec ed e o a e: looking a , ins ead o away om, he a ge
‑An icipa ion a e: making a saccade du ing he ixa ion pe iod, p io o he p es‑
en a ion o he a ge do o wi hin 80 ms o i s p esen a ion
Memo y‑guided saccade Memo y‑guided saccade Gain, la ency, and peak eloci y o co ec saccades E o s ‑Inhibi o y e o s: Looking a he do when i was p esen ed o making a saccade
be o e he esponse pe iod
‑Di ec ional e o s: Looking in he w ong di ec ion o whe e he s imulus had been
p esen ed du ing he esponse ime
Page 7 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
comp ised o 55 pa icipan s, wi h 29 in he “Con ol”
g oup and 26 in he “Subclinical” g oup.
Pa icipan s
Table 2 shows he desc ip i e s a is ics o he pa ici-
pan s’ sociodemog aphic a iables. The demog aphic
a iables age and BMI e idenced no s a is ically signi i-
can di e ences be ween he “Con ol” and “Subclinical”
g oups, while all psychological measu es used o spli -
ing he sample showed signi ican di e ences, wi h sig-
ni icance alues less han 0.001 o mos compa isons.
E ec sizes anged om mode a e o high. Unexpec edly,
bo h g oups di e ed om each o he in e ms o anxi-
e y, wi h he con ol g oup sco ing highe han he sub-
clinical g oup. In o de o exclude any po en ial e ec o
anxie y on ou esul s, we compa ed he same a iables
desc ibed in he S a is ical Analysis’ sec ion abo e, bu
including anxie y as a co a ia e (ANCOVAs). All he
analyses showed ha anxie y did no ha e any e ec in
a ying ou esul s. The e o e, o he sake o simplici y,
we epo - es s’ ou comes in he main ex .
Fixa ion ask
No signi ican di e ences (p > 0.005) we e ound be ween
he “Con ol” g oup (M = 1.688; SD = 3.856) and he
“Subclinical” g oup (M = 1.385; SD = 2.082) in e ms o
he numbe o saccadic in usions pe o med pe minu e.
P osaccade/an isaccade ask
In he p osaccade ask (see Table3), gain and la ency
be ween he “Con ol” and “Subclinical” g oups showed
no s a is ically signi ican di e ences (p > 0.005). Peak
eloci y and an icipa ion a e also e ealed no signi i-
can di e ences be ween g oups (p > 0.005). P osaccade
e o s showed s a is ically signi ican di e ences, wi h
a low e ec size (p = 0.034). Ne e heless, a e applying
he False Disco e y Ra e (FDR) co ec ion using he Ben-
jamini–Hochbe g p ocedu e ac oss he 23 compa isons,
he esul s we e non-signi ican . Gi en he c i e ia se by
Table 2 Demog aphic and psychological cha ac e is ics
n 29 “Con ol”; n = 26 “Subclinical”
a S uden ’s
b Mann–Whi ney U
Va iables M (SD)
Con ol M (SD)
Subclinical S a is ic pE ec size
Age 20.310
(1.417) 20.038
(1.428) 436.500 b.308 .158
BMI 21.285 (1.578) 21.769
(1.889) ‑1.036 a.305 ‑.280
STAIT‑ S a e 49.714
(7.605 38.880
(9.369) 568.000 b < .001 .623
STAIT‑ T ai 40.607
(11.364) 31.120
(10.764) 530.000 b.001 .514
P emo bid in elligence 8.214
(3.425) 8.182
(2.612) .037 a.971 .010
S‑EDE‑Q Res ain .593
(.514) 2.462
(1.118) ‑7.812 b < .001 ‑2.147
S‑EDE‑Q Ea ing conce n .345
(.362) 1.438
(1.042) 75.500 b < .001 ‑.800
S‑EDE‑Q Shape conce n .841
(.576) 3.288
(1.256) ‑9.452 a < .001 ‑2.553
S‑EDE‑Q Weigh conce n .434
(.421) 2.731
(1.243) 3.000 b < .001 ‑.992
S‑EDE‑Q Global .553
(.310) 2.480
(.951) ‑10.324 a < .001 ‑2.788
Table 3 P osaccadic ask esul s
n = 28 “Con ol”; n = 26 “Subclinical”
a = S uden ’s
b = Mann–Whi ney U
Va iables M (SD)
Con ol M (SD)
Subclinical S a is ic pE ec size
Gain .981
(.055) .988
(.036) ‑0.562 a.577 ‑.153
La ency 317.090
(62.814) 315.330
(91.580) 416.000 b.376 .143
Peak eloci y 303.453
(45.338) 319.380
(51.128) ‑1.213 a.231 ‑.330
An icipa ion a e .569
(.095) .544
(.069) 1.128 a.264 .307
P osaccade e o s .00
(.004) .006
(.012) 292.000 b.034 ‑.198
Page 8 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
his co ec ion me hod, only he mos signi ican ind-
ings, pa icula ly hose wi h p- alues no ably less han
0.001, emain s a is ically signi ican . This adjus men
p ocess e ealed ha he p e iously no ed signi ican di -
e ence in p osaccade e o s, despi e i s low e ec size,
did no wi hs and he FDR adjus men .
In he an isaccade ask (see Table4), all compa isons
be ween he “Con ol” and “Subclinical” g oups, includ-
ing accu acy, la ency, peak eloci y, an icipa ion a e,
unco ec ed e o a e, co ec ed e o a e, and la ency
o he co ec ed saccade, showed no s a is ically signi i-
can di e ences. E ec sizes we e low, indica ing ha ,
indeed, di e ences be ween g oups on hese ocula
measu es a e minimal.
Memo y‑guided saccade ask
In he memo y-guided saccade ask (see Table5), com-
pa isons be ween he “Con ol” and “Subclinical" g oups
in e ms o gain a bo h 5° and 10°, la ency a 5° and 10°,
peak eloci y a 5° and 10°, as well as inhibi o y e o s
and di ec ional e o s a 5° and 10°, e ealed no s a is-
ically signi ican di e ences. E ec sizes we e low o
all measu es collec ed, indica ing minimal di e ences
be ween g oups. No ably, analysis was no pe o med o
inhibi o y e o s a 5° because he a iance in hese da a,
a e g ouping by “Subclinical”, was equal o ze o.
Explo a o y co ela ional analysis
No signi ican co ela ions we e ound be ween he
di e en measu es o eye mo emen s and he ED symp oma ology a iables measu ed by he S-EDE-Q
ques ionnai e. The absence o signi ican di e ences p e-
cluded he disc iminan analysis de ailed in he p o ocol.
Table 4 An isaccade ask esul s
n = 28 “Con ol”; n = 26 “Subclinical”
a = S uden ’s
b = Mann–Whi ney U
Va iables M (SD)
Con ol M (SD)
Subclinical S a is ic pE ec size
Gain 1.046
(.230) 1.005
(.254) .627 a.533 .171
La ency 437.953
(100.787) 446.210
(112.458) 345.000 b.751 ‑.052
Peak eloci y 281.549
(55.053) 280.671
(64.440) .054 a.957 .015
An icipa ion a e .342
(.156) .328
(.133) 376.500 b.835 .034
Unco ec ed e o a e .169
(.125) .173
(.128) 367.000 b.965 .008
Co ec ed e o a e .144
(.114) .169
(.138) 327.000 b.530 ‑.102
La ency o he co ec ed saccade 540.904
(165.400) 528.069
(127.210) 410.000b.434 .126
Table 5 Memo y‑guided saccade esul s
n = 29 “Con ol”; n = 26 “Subclinical”
a = S uden ’s = S uden ’s
b = Mann–Whi ney U
c = Inhibi o y E o s 5º analysis was no pe o med because he a iance in he
da a, a e clus e ing acco ding o he “G oup” ac o , was equal o 0
Va iables M (SD) M (SD) S a is ic pE ec size
Con ol Subclinical
Gain 5º .962
(.413) .862
(.119) 390.000b.834 .034
Gain 10º .954
(.131) .955
(.099) 351.000 b.670 ‑.069
La ency 5º 705.859
(199.840) 703.792
(217.105) 387.000 b.874 .027
La ency 10º 674.987
(162.821) 658.272
(184.720) 437.000 b.319 .159
Peak eloci y 5º 210.309
(42.224) 205.640
(37.169) 394.000 b.783 .045
Peak eloci y 10º 278.263
(58.723) 282.047
(56.044) ‑.244 b.808 ‑.029
Inhibi o y e o s
5º .000
(.00) .002
(.009) ‑ c‑ ‑
Inhibi o y e o s
10º .023
(.037) .013
(.024) 427.000 b.316 .133
Di ec ional e o s
5º .010
(.022) .012
(.034) 391.500 b.726 .038
Di ec ional e o s
10º .011
(.022) .012
(.025) 372.00 b.919 ‑.013
Page 9 o 12
Na as‑Leóne al. BMC Psychia y (2025) 25:133
Discussion
The goal o his esea ch was o explo e he exis ence
o eye mo emen changes, p e iously iden i ied in indi-
iduals wi h AN, wi hin a popula ion wi h subclini-
cal ED symp oma ology. P io esea ch has highligh ed
hose indi iduals wi h AN exhibi an inc eased numbe
o SWJs [13, 14] and a highe numbe o inhibi o y con-
ol e o s [15, 16]. Those changes ha e been p oposed
as digi al bioma ke s (measu ed by he ixa ion ask)
and endopheno ypes (measu ed by he p osaccade/an i-
saccade and memo y-guided asks) [5]. Such changes in
eye mo emen pa e ns may al eady mani es in people
wi h subclinical ED symp oma ology, which would open
po en ial u u e esea ch di ec ions o explo e he pos-
sibili ies o his echnique o ea ly de ec ion and p e-
en ion. Howe e , un il now he e ha e been no s udies
making ou s udy a no el con ibu ion o his a ea o
esea ch. To achie e his, we assessed pa icipan s’ pe -
o mance on a se ies o saccadic eye mo emen asks,
including ixa ion, p osaccade, an isaccade, and memo y-
guided saccade asks. We hypo hesized ha pa icipan s
wi h subclinical ED symp oma ology compa ed o con-
ols would be p one o show sho e saccade la encies
and mo e inhibi o y e o s in a p osaccade/an isaccade
ask and in a memo y-guided saccade ask, espec i ely,
and an inc eased a e o SWJs in a ixa ion ask.
Howe e , and con a y o ou ini ial hypo hesis [6],
he indings indica e ha lowe -o de isual p ocessing
emains well-p ese ed among indi iduals wi h subclini-
cal ED symp oma ology. This is somehow consis en wi h
exis ing li e a u e, which sugges s ha indi iduals wi h
AN pe o med a ela i ely “compa able” pe o mance o
con ol pa icipan s on hese eye mo emen asks, excep
o he speci ic changes p oposed abo e (i.e., inhibi o y
e o s) [15, 16]. This is also in line wi h Beilha z e al.
[42], who ound no signi ican di e ences in saccadic eye
mo emen ask pe o mance be ween indi iduals wi h
body dysmo phic diso de and a con ol g oup, aside
om a signi ican end o inc eased an icipa o y e o s
in he p osaccade ask among he o me g oup [42].
Phillipou e al. [16] obse ed, using p osaccade/an isac-
cade and memo y-guided asks, ha indi iduals wi h AN
commi ed only mo e inhibi o y e o s han hose in he
weigh es o ed AN g oup, sis e s o indi iduals wi h AN,
and heal hy con ols. They sugges ed ha hese changes
migh be a speci ic endopheno ype o he acu e phase o
he illness, e lec ing unde lying neu obiological impai -
men s. Building on hese indings, i would be wo h
in es iga ing whe he an a oidance s a egy, simila o
a en ional biases owa d ood s imuli in clinical samples
[43], could be obse ed in indi iduals wi h subclinical ED
symp oms. Adap ing he p osaccade ask wi h ood cues
migh help explo e his possibili y.
In addi ion, Phillipou e  al. [13] and Phillipou e  al.
[14] p oposed saccadic in usions (measu ed by he ixa-
ion ask) as a po en ial bioma ke (i.e., a ai ). Acco d-
ing o hei indings, pa ien s wi h AN, hose who had
eco e ed om he illness, and sis e s o people wi h AN
exhibi ed signi ican ly mo e in usions han heal hy con-
ols, in con as o ou esul s. Howe e , as p e iously
discussed, i is c ucial o acknowledge he limi a ions o
ou s udy be o e d awing de ini i e conclusions. These
limi a ions encompass he he e ogenei y o he sample
wi h subclinical ED symp oma ology and he a iabil-
i y in he mani es a ion o subclinical ED symp oma ol-
ogy. Conside ing his pe spec i e, and as sugges ed by
Phillipou e  al. [14], he complexi y o he p oposed
bioma ke —shaped by gene ic, neu obiological, and
en i onmen al ac o s—sugges s ha i s p esence migh
no mani es uni o mly ac oss di e se popula ions wi h
subclinical ED symp oma ology.
Gi en he indings, we p opose ha he isual p ocess-
ing changes iden i ied in p e ious s udies may be closely
associa ed wi h he ac i e clinical phase o he illness and
may no s a o mani es in a he e ogeneous popula-
ion wi h subclinical ED symp oma ology. F om a neu-
obiological pe spec i e, his sugges s a well-p ese ed
unc ioning in b ain egions associa ed wi h SWJs and
olun a y saccade con ol. These esul s imply ha such
bioma ke s (i.e., saccadic in usions) and endopheno-
ypes (i.e., inhibi o y e o s) migh no be cha ac e is-
ic ai s o he e ogeneous EDs samples wi h subclinical
symp oma ology.
Implica ions o p e en i e s a egies inea ing diso de s
Ea ly iden i ica ion o indi iduals wi h subclinical ED
symp oma ology is challenging bu is c i ical o he
imely implemen a ion o in e en ions and he e icacy
o ea men s a egies [44]. Conside ing he e idence
suppo ing s a e-dependen eye mo emen changes du -
ing he ac i e phase o he illness [16], he p ese a ion o
lowe -o de isual p ocessing—indica ing well-p ese ed
cogni i e unc ions ela ed o inhibi o y con ol and
isual memo y—could aid in di e en ia ing indi iduals
wi h subclinical ED symp oma ology o in a heal hy s a e
om hose wi h AN. This app oach aligns wi h ecen
ad ances in he use o digi al bioma ke s in psychia ic
esea ch, o e ing new oppo uni ies o ea ly de ec ion
and pe sonalized in e en ions h ough senso echnolo-
gies such as eye acking [45]. Fo example, se e al s ud-
ies emphasize he po en ial o u ilizing web-cam based
eye acke s, which emo e he necessi y o cos ly equip-
men , a el, and specialized pe sonnel, making i easie
o conduc egula and long- e m assessmen s o eye
mo emen s mo e accessible and easible o in eg a ion