Social Cogni i e and A ec i e Neu oscience, 2025, 20(1), nsa 114
DOI: h ps://doi.o g/10.1093/scan/nsa 114
Ad ance Access Publica ion Da e: 31 Oc obe 2025
O iginal Resea ch – Neu oscience
Beyond uni o mi y: indi idual sensi i i ies o ewa d and
punishmen shape mid on al- he a esponses o
app oach a oidance con lic
Shubham Pandey*, and Roman Osinsky
Di e en ial Psychology and Pe sonali y Resea ch Uni , Ins i u e o Psychology, Osnab ück Uni e si y, Osnab ück 49076, Ge many
*Co esponding au ho . 75/236, Di e en ial Psychology and Pe sonali y Resea ch Uni , Ins i u e o Psychology, Osnab ück Uni e si y, Lise-Mei ne -S . 3 49076
Osnab ück, Ge many. E-mail: [email p o ec ed]
Abs ac
App oach-a oidance con lic (AAC) is a co e aspec o decision-making, in ol ing compe ing appe i i e and a e si e ou comes. Gi en
subs an ial indi idual di e ences in sensi i i y o ewa d and punishmen , AAC expe iences likely a y ac oss indi iduals. This s udy
examined he neu ocogni i e mechanisms unde lying AAC, ocusing on hese indi idual di e ences. Pa icipan s comple ed a ask
wi h ou le els o ewa d and punishmen p obabili ies (.25, .50, .75, 1), choosing o accep o ejec ewa ds pai ed wi h po en ial
punishmen s. Impo an ly, we did no ind a uni o m condi ion ha consis en ly elici ed maximum o minimum AAC ac oss pa ici-
pan s. We iden i ied indi idualized high- and low-con lic condi ions using h ee me ics: eac ion ime, ewa d ejec ion a e, and a
composi e Beha iou al Con lic Index. Mid- on al he a (MFT) powe , an es ablished ma ke o con lic moni o ing, was signi ican ly
highe in high-con lic compa ed o low-con lic ials. Compu a ional modelling e ealed a iabili y in pa icipan s’ weigh ing o ewa d
and punishmen p obabili ies, e lec ing di e ences in pa icipan s’ sensi i i y o ewa d and punishmen ou comes. Fu he mo e,
sel - epo ed Beha iou al Inhibi ion Sys em sco es p edic ed MFT esponses, linking pe sonali y ai s o AAC p ocessing. These indings
demons a e ha MFT powe e lec s AAC-d i en cogni i e con ol, shaped by indi idual sensi i i ies and ai s. Ou esul s emphasize
he impo ance o pe sonalized con lic de ini ions o unde s anding adap i e con ol in mo i a ionally ambiguous con ex s.
Keywo ds: EEG; mid- on al he a; app oach; a oidance; cogni i e con ol
In oduc ion
Cogni i e con ol and con lic moni o ing a e undamen al p o-
cesses ha enable indi iduals o na iga e compe ing mo i a ional
demands in e e yday decision-making. One c ucial o m o con lic
a ises when indi iduals mus esol e he ension be ween concu -
en ly ac i a ed app oach and a oidance endencies. Tha is, he
same beha iou in ol ed in ob aining po en ial ewa d also po en-
ially leads o an a e si e ou come. Unde s anding how he b ain
manages app oach-a oidance con lic s (AAC) has b oad implica-
ions, anging om heo ies o decision-making o applica ions in
clinical se ings o diso de s ma ked by impai ed cogni i e con ol,
such as anxie y and addic ion (Le kiewicz e al. 2023).
Mid- on al he a (MFT) ac i i y, o igina ing om he pos e io
mid on al co ex (pMFC), has been consis en ly implica ed in
cogni i e con ol and con lic moni o ing (Ca anagh e al. 2009,
Cohen and Donne 2013, Ca anagh and F ank 2014, Dup ez e al.
2020, Mu alidha an e al. 2023). In pa icula , ansien inc eases
in MFT powe a e hough o e lec neu al mechanisms esponsi-
ble o de ec ing con lic , signalling he need o con ol, and acil-
i a ing adap i e beha iou al esponses. Despi e ex ensi e esea ch
on MFT in simple o ms o s imulus- esponse con lic asks,
esea ches ha e jus begun o in es iga e he ole o MFT in he
con ex o AAC, whe e mo i a ional and a ec i e p ocesses s ongly
in luence beha iou (Ullspe ge e al. 2014, Lange e al. 2022, Ziebell
e al. 2023).
Mo eo e , p e ious s udies on con lic p ocessing ha e mainly
elied on p e-assumed uni o m “high con lic ” and “low con lic ”
classi ica ion (e.g. incong uen and cong uen ials in s oop/
lanke asks). Howe e , ecen indings sugges ha an indi idu-
alized app oach can be e in o m us abou neu ocogni i e mech-
anisms o con lic p ocessing. Fo ins ance, Pinne and Ca anagh
(2017) showed ha MFT powe can accoun o a ying beha iou al
endencies ac oss indi iduals. In ano he s udy, Lin e al. (2018)
showed ha indi idualized con lic pa ame ically modula ed MFT
powe . Thei s udy ailo ed con lic le els o each pa icipan using
pa ame e s ex ac ed om an in e empo al choice ask ha pa -
icipan s pe o med be o e he Elec oencephalog am (EEG) ask.
Wi h ega d o AAC, Lange e al. (2022) ound sys ema ic indi idual
di e ences in ewa d-punishmen con igu a ions ha led o high-
es le el o AAC and co esponding MFT esponses. I is e y likely
ha such di e ences a ise om a iabili y in s able sensi i i ies
o ewa d and punishmen cues and, consequen ly, app oach-
a oidance endencies (G ay and McNaugh on 2000, Co 2004).
Consequen ly, he poin o maximal AAC is p obably no uni o m
Recei ed: 14 May 2025; Re ised: 17 Sep embe 2025; Accep ed: 24 Oc obe 2025
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 (h ps://c ea i ecommons.o g/licenses/by/4.0/), which
pe mi s un es ic ed euse, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed.
© The Au ho (s) 2025. Published by Ox o d Uni e si y P ess.
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026
2 | Pandey and Osinsky
ac oss indi iduals. Neglec ing such indi idual di e ences in AAC
p ocessing may signi ican ly diminish con as s a is ics o mea-
su es o neu al ac i i y (compa e, Ahn e al. 2011). In he p esen
s udy, we he e o e in es iga ed he link be ween MFT and
AAC-p ocessing by inco po a ing indi idualized de ini ions o con-
lic condi ions, he eby accoun ing o pa icipan s’ unique sensi-
i i ies o ewa ds and punishmen . We use a ask wi h mul iple
combina ions o ewa d and punishmen p obabili ies o be e
accoun o in e -indi idual a iabili y in sensi i i y o ein o ce s.
Ra he han elying on p e-assumed “high con lic ” and “low con-
lic ” condi ions (e.g. incongu en and cong uen ials in S oop
asks o simila asks, ixed combina ion o ewa d and punishmen
he e), we ope a ionalized con lic on a pa icipan -by-pa icipan
basis using beha iou al con lic -indices such as eac ion ime and
ewa d ejec ion a e.
Me hod
Pa icipan s, appa a us and s imuli
Fo y olun ee s (29 emales) aged 18 o 32 yea s (M = 22.05 yea s,
SD = 3.07 yea s) wi h no mal o co ec ed no mal ision ook pa
in his s udy. All pa icipan s we e ec ui ed h ough lye ad e -
isemen s, p o ided hei in o med w i en consen , and epo ed
being in good heal h, ee o medica ions, and wi hou any his o y
o psychia ic o neu ological disease.
The s udy was implemen ed using he PsyToolbox in MATLAB
(Ma hWo ks Inc.). The expe imen al session was conduc ed in a
dimly li oom, wi h pa icipan s sea ed a a dis ance o ∼90 cm in
on o a 24-inch LCD la -sc een moni o (144 Hz e esh a e and
1920 × 1080 esolu ion). The expe imen al p o ocol began wi h a
b ie su ey collec ing pa icipan s’ biog aphic de ails ollowed by
he main beha iou al ask.
In each ial, wo e ical ba s o same size we e p esen ed on
sc een (see Fig. 1): a ed o ewa d, a g een o punishmen . A black
dash ma ke posi ioned on each ba indica ed he p obabili y o
ewa d and punishmen . The black dash could appea a one o he
ou equidis an posi ions, co esponding o p obabili ies .25, .50,
.75, o 1, esul ing in 4 × 4 = 16 ewa d-punishmen p obabili y com-
bina ions. Pa icipan s we e in o med ha hey could win poin s,
which hey can la e con e in o mone a y ewa d o a ound en
Eu o. As a punishmen , pa icipan s hea d one o h ee a e si e
sounds (90 dB) deli e ed h ough an o e - he-ea headphone, an-
domly selec ed.
Expe imen p ocedu e
The sc een backg ound colou was se o g ey [0.7 0.7 0.7]. Each
ial began wi h a cen ally p esen ed ixa ion las ing 500 ms. This
was ollowed by p esen a ion o wo e ical ba s: a ed one and a
g een one, appea ing on le and igh side o sc een, coun e bal-
anced ac oss pa icipan s, o a du a ion o 1500 ms and equidis an
om cen e o sc een (2.7°). We e e he g een ba as ewa d me e
and he ed ba as punishmen ( h ea ) me e . Pa icipan s we e
ins uc ed o quickly p ess le o igh a ow key bu on o accep
o decline he ewa d. The side o he ewa d s. punishmen me e ,
as well as he mapping o he accep s. decline bu ons, was always
aligned and coun e balanced ac oss pa icipan s. The maximum
ime o esponse was same as he ime o s imulus display, and he
s imulus s ayed on sc een o en i e du a ion o 1500 ms i espec-
i e o pa icipan esponse.
S imulus display was ollowed by a eedback o 700 ms whe e
a box was shown. The e we e h ee possible pa icipan esponses:
accep he ewa d, decline he ewa d, o make no esponse wi hin
he allo ed ime. I he pa icipan accep ed he ewa d, one o
ou ou comes could occu based on he p ede ined p obabili ies:
(1) a sole ewa d, (2) a sole punishmen , (3) bo h ewa d and pun-
ishmen , o (4) nei he ewa d no punishmen . In a ewa d only
ial, hey we e shown a g een box wi h “+5 poin s” supe imposed
on i . In a bo h ewa ding and punishing ial, hey saw a hal
ed-hal g een box wi h “+5 poin s.” Punishmen -only ials we e
indica ed by a ed box. T ials wi h nei he ewa d no punishmen
yielded no eedback. The po en ial ou come o a gi en ial was
p ede ined, aking he ewa d and punishmen p obabili ies in o
accoun . Winning a ewa d was solely de e mined by ewa d p ob-
abili y, e.g. a ewa d p obabili y o 1 led o ewa d on all such ials,
Figu e 1. App oach a oidance ask. The ho izon al line on he ba s ep esen s he p obabili y (.25, .50, .75, 1) o ewa d (g een ba ) and punishmen
( edba ) (Ge man: Belohnung: ewa d; Bes a ung: punishmen ). In each ial, pa icipan needed o decide whe he o go o he ewa d unde he
gi en cons ella ion o ewa d-punishmen p obabili ies.
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026
App oach-a oidance con lic | 3
while a ewa d p obabili y o .5 led o a ewa d on only 50% o such
ials. The same p inciple applied o punishmen p obabili y and
ac ual punishmen deli e y. In case pa icipan s declined he
ewa d, hey ecei ed nei he ewa d no punishmen , as indica ed
by a black box. I no esponse was made wi hin he esponse win-
dow, pa icipan s o ei ed he ewa d bu could s ill ecei e pun-
ishmen ( ed box), depending on he p ede e mined ial ou come.
I such a ial did no con ain punishmen , no eedback was shown.
Following he eedback, he e was an in e - ial in e al o mean
1500 ms d awn om a pseudo-Gaussian dis ibu ion anging om
1000 ms o 2000 ms.
The expe imen consis ed o eigh blocks, each comp ising o 64
ials. Each o he six een ewa d-punishmen p obabili y combi-
na ions was ep esen ed by ou ials in a block, o alling 32 ials
pe condi ion ac oss he expe imen . T ials p esen a ion was
pseudo- andomized o ensu e ha no condi ion was epea ed in
subsequen ials. A e each block, pa icipan s we e shown hei
cumula i e ewa d poin s. Each pa icipan unde wen an ini ial
p ac ice session o 4 ials o each o six een condi ions o amil-
ia ize hemsel es wi h he ask. The ask oughly ook 40 minu es
o comple e, wi h an addi ional 60 minu es o EEG p epa a ion and
hai wash.
Resea ch design and da a analysis
To add ess ou main esea ch ques ion, we aimed o con as con-
di ions o maximum con lic wi h hose o minimum con lic . We
implemen ed wo dis inc app oaches o de ining hese condi ions,
as desc ibed in de ail below.
In he uni o m app oach we assumed he same ask condi ion
de ini ions ac oss all pa icipan s. Speci ically, we posi ed ha he
50% ewa d—50% punishmen condi ion would elici highes
whe eas he 100% ewa d—25% punishmen o 25% ewa d—100%
punishmen condi ion would ep esen lowes AAC.
In he indi idualized app oach, we accoun ed o in e -indi idual
a iabili y by iden i ying pe sonalized condi ions o maximum and
minimum con lic . These we e de e mined based on pa icipan ’s
ewa d ejec ion a es, eac ion imes (RTs), o a combina ion o
bo h. In each me hod, we selec ed he op h ee and bo om h ee
condi ions o ep esen high and low AAC, espec i ely. We picked
h ee condi ions ins ead o one condi ion o wo easons: i s , i
inc eased o al a ailable ials o EEG analysis, and second, mo e
impo an ly, o se e al pa icipan s he e we e mo e han one
condi ion wi h 0% o 100% ewa d ejec ion a e. Fo he sole RT
based app oach, we assumed he h ee condi ions wi h highes
mean RT as maximum con lic condi ion, and he h ee condi ion
wi h lowes RT as minimum con lic condi ions. Fo he sole
ejec ion- a e based app oach, we assumed he h ee condi ions
wi h ewa d ejec ion- a e closes o 50% as highes con lic con-
di ions, and he h ee condi ions wi h ejec ion- a e closes o 0 o
100% as minimum con lic condi ions. Howe e , we ound ha , o
some pa icipan s, he e we e mo e han h ee condi ions wi h
same ewa d ejec ion a e (e.g. 0% o all o condi ions whene e
punishmen p obabili y was 0.25). Fo hese cases, we andomly
chose h ee condi ions ou o all quali ying condi ions.
In ou hi d indi idualized app oach, we combined RT and
ewa d ejec ion a e o c ea e a new me ic e lec ing he indi id-
ual deg ee o expe ienced con lic . We used his me ic o iden i y
indi idualized condi ions o maximum and minimum con lic . We
ook his app oach based on p e ious indings, indica ing he ha
inclusion o eac ion imes along wi h choice da a ( ejec ion a es)
imp o e he accu acy o me ic es ima ion (P e au e al. 2009, Bal-
la d and McClu e 2019, Zo owi z e al. 2019). No ice ha ejec ion
a e exhibi s a V-shaped dis ibu ion, wi h maximum con lic
occu ing a ound 50% accep ance/ ejec ion a e. Con lic inc eases
as he ejec ion a e mo es om ei he ex eme (0% o 100%)
owa d 50%. The e o e, we ans o med ejec ion a e in o a linea
me ic, called de ia ion using he o mula.
De ia ion Rewa d ejec ion a e
=−−05 05..
The de ia ion sco e anges om 0 (minimal con lic ) o 0.5 (max-
imal con lic ). Acco dingly, bo h RT and de ia ion sco es a e expec ed
o linea ly ela e o indi idualized con lic . In he nex s ep, we
combined hese wo me ics, and compu ed a composi e sco e,
named “Beha iou al Con lic Index” as ollows:
Be a iou al
h Con lic Index BCI Reac ion ime De ia i
()
= × oo n +
()
ε
whe e ε was se o 0.01 o p e en beha iou al con lic index (BCI)
becoming 0 in case de ia ion was 0 (i.e. when ewa d ejec ion a e
was 0% o 100%). This way, we calcula ed BCI sco es o each o he
16 AAC condi ions o each pa icipan and picked condi ions wi h
maximum and minimum sco es.
All analyses we e pe o med in Ma lab (The Ma hWo ks, USA,
e sion 2023 b) using cus om w i en code. S a is ical es s we e
pe o med in JASP so wa e (Lo e e al. 2019). We epo pa ial e a
squa e (ηp
2), and Cohen’s d (d, mean o he di e ence sco es di ided
by hei s anda d de ia ion) as e ec size measu es. Fo co ela ion,
we epo Pea son co ela ion coe icien s, . Mean and s anda d
de ia ion a e epo ed wi h usual no a ion as M and SD espec i ely.
Ques ionnai e
Be o e he s a o expe imen , pa icipan s comple ed he Ge man
e sion o Rein o cemen Sensi i i y Theo y o Pe sonali y Ques-
ionnai e (RST-PQ; Co and Coope 2016, Pugnaghi e al. 2018).
Rein o cemen Sensi i i y Theo y (G ay and McNaugh on 2000,
Co 2004) posi s h ee majo neu obiological sys ems: he
Beha iou al App oach Sys em (BAS), which p omo es app oach
beha iou owa d appe i i e s imuli; he Figh –Fligh –F eeze Sys-
em (FFFS), which acili a es ac i e a oidance o a e si e s imuli;
and he Beha iou al Inhibi ion Sys em (BIS), which ac s as a supe -
iso y con lic -de ec ion mechanism. The BIS is conside ed o be
supe o dina e o bo h BAS and FFFS and is speci ically ac i a ed
by mo i a ional con lic s, especially hose a ising om simul ane-
ous and opposing app oach and a oidance endencies. Based on
his amewo k, we hypo hesized ha BIS sco es would posi i ely
co ela e wi h mid- on al he a (MFT) powe in esponse o AAC.
EEG da a eco ding and p ocessing
EEG da a we e eco ded wi h a 500 Hz sampling a e and a 250 Hz
low-pass and 0.016 Hz high-pass il e , using he 64 channel s an-
da d ac iCAP sys em, B ainAmp Ampli ie s and B ainVision
Reco de so wa e (all B ain P oduc s, Gilching, Ge many). The e -
e ence elec ode and he g ound elec ode we e placed a posi ions
FCz and AFz, espec i ely. O -line p ocessing was conduc ed wi h
B ainVision Analyze 2 so wa e. An independen componen anal-
ysis (ICA) was applied o de ec ion and emo al o ocula a e ac s.
EEG da a we e hen e- e e enced o he a e age o all elec odes,
and he o me e e ence a FCz was eins a ed as a new da a
channel. The con inuous EEG da a we e sepa a ed in o 2500 ms
segmen s a ound he onse o he ba s-s imulus (−1000 o 1500 ms).
T ials exhibi ing ol age s eps >50 µV/ms, di e ences >400 µV
wi hin 600 ms, o low ac i i y de ined by < 0.5 µV wi hin 100 ms
we e ejec ed om u he analyses. The emaining ials we e
used o he ime- equency analysis and o he Gene alized
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026
4 | Pandey and Osinsky
Eigen-Decomposi ion (GED; Cohen 2022), conduc ed using MATLAB
R2020a (Ma hWo ks).
Time- equency analysis
Fo he ime- equency analysis, a wa ele ans o ma ion was
applied on single- ial da a using 30 complex Mo le wa ele s,
ei2π e- 2/(2s2), whe e is ime, is equency ( anging om 1 o 30 Hz
in 30 loga i hmically spaced s eps), and s is he wid h o each e-
quency band, which is de ined by n/(2π ) (wi h n loga i hmically
inc easing om 4 o 7). F equency speci ic powe a each ime poin
( ) was de ined as he squa ed magni ude o he esul ing analy ic
signal (Z) as eal [Z( )2] + imagina y [Z( )2]. Resul ing powe le els
we e dB-baseline co ec ed pe equency laye using he a e age
powe o all ials wi hin he −800 o −300 ms baseline ime window.
We conduc ed clus e -based pe mu a ion es s (Ma is and Oos en-
eld 2007). We used di e ence o maximum and minimum con lic
condi ion da a om channel FCz, clus e - o ming h esholds o
P = .001, compu ed 1000 i e a ions. This yielded a signi ican egion
on ime- equency plo wi h an app oxima e ime window om
500 o 1100 ms. MFT was ex ac ed as mean powe wi hin 4 Hz o
8 Hz a FCz.
Gene alized Eigen-decomposi ion and sou ce econs uc ion
We pe o med GED o sepa a e di e en sou ces o he a band
ac i i y and hen singled ou componen s showing a mid on al
opog aphy. This p ocedu e is based on he app oach aken by
(Zuu e e al. 2020). The GED was designed o yield il e s ha max-
imally di e en ia e be ween b oadband EEG ac i i y and he a
band ac i i y. The e o e, un il e ed bu z-sco ed single- ial EEG
ac i i y was used o compu ing a e e ence ma ix, whe eas he
signal ma ix was compu ed using band-pass- il e ed (4–8 Hz) and
z-sco ed single- ial ac i i y. Bo h ma ices a e based on he ime
window o 0–1000 ms a ound s imulus onse . Gi en he 65 da a
channels, he GED ex ac ed 65 componen s pe pa icipan . The
signi icance o hese componen s was assessed using pe mu a ion
es s. A null dis ibu ion o eigen alues was gene a ed by shu ling
he a band il e ed and un il e ed b oadband ac i i y ime se ies
1000 imes. To co ec o mul iple compa isons, he 95 h pe cen ile
o all gene a ed eigen alues was used o signi icance es ing ( ed
line in Fig. 2) (Hay on e al. 2004). Componen s whose eigen alues
we e no signi ican ly g ea e han he eigen alues om he an-
domly shu led da a (no exceeding a h eshold o α = 0.05) we e
excluded om u he analyses.
Nex , opog aphical maps we e compu ed o each componen .
These opog aphies we e compa ed wi h a p o o ypical empla e
(Gaussian cen ed a FCz) e lec ing mid on al opog aphical ac i -
i y (Zuu e e al. 2020). Componen s exhibi ing sha ed spa ial a i-
ance wi h he mid on al empla e less han R2 < 0.6 we e excluded
om u he analysis. Las , he ime se ies ac i i y o each compo-
nen was compu ed by mul iplying he eigen ec o wi h he EEG
ime se ies ac i i y and hen applying a ime- equency decompo-
si ion using he same pa ame e s o a wa ele ans o ma ion as
desc ibed o he p e ious ime- equency analysis. This p ocedu e
ex ac ed a inal se o GED componen s and hei ime- equency
ac i i y o each pa icipan (Zuu e e al. 2020). S a is ical analyses
we e pe o med based on he esul ing componen ime- equency
ac i i ies.
Resul
Beha iou al esul s
Reac ion ime and ewa d ejec ion a e
Fo RT, ejec ion a es and BCI sco es, we pe o med sepa a e
wo-way epea ed measu e ANOVAs wi h he wo ac o s ewa d
p obabili y (.25, .5, .75, 1) and punishmen p obabili y (.25, .5, .75, 1).
Fo RT we obse ed a main e ec o ewa d p obabili y, F (3, 117) =
43.22, p < .001, ηp
2 = 0.52; a main e ec o punishmen p obabili y,
F (3, 117) = 18.59, p < .001, ηp
2 = 0.32; and an in e ac ion e ec o
bo h ac o s, F (9, 324) = 5.84, p < .001, ηp
2 = 0.13 (Fig. 3). Simila ly,
o ewa d ejec ion a e we obse ed a main e ec o ewa d p ob-
abili y, F (3, 117) = 145.62, p < .001, ηp
2 = 0.79; a main e ec o pun-
ishmen p obabili y, F (3, 117) = 128.95, p < .001, ηp
2 = 0.77; and an
Figu e 2. Signi ican componen s iden i ied wi h GED o one model pa icipan .
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026
App oach-a oidance con lic | 5
in e ac ion e ec o bo h ac o s, F (9, 324) = 7.36, p < .001, ηp
2 = 0.15.
Consequen ly, o BCI sco es we also obse ed a main e ec o
ewa d p obabili y, F (3, 117) = 11.93, p < .001, ηp
2 = 0.23; a main e ec
o punishmen p obabili y, F (3, 117) = 6.06, p < .001, ηp
2 = 0.13; and
an in e ac ion e ec o bo h ac o s, F (9, 324) = 2.87, p = .003,
ηp
2 = 0.06.
The main e ec s e ealed ha highe ewa d p obabili ies we e
associa ed wi h as e eac ion imes and lowe ejec ion a es,
whe eas highe punishmen p obabili ies we e associa ed wi h
slowe eac ion imes and highe ejec ion a es. In o he wo ds,
he likelihood o winning poin s acili a ed as e and mo e accep -
ing decisions, while he likelihood o punishmen induced slowe
and mo e ejec ing decisions. Gi en he ela i ely la ge numbe o
condi ions and ac o -le el combina ions, in e p e ing he in e ac-
ion e ms was less s aigh o wa d. Conduc ing exhaus i e pai -
wise compa isons (120 in o al) would be s a is ically imp ac ical.
Sepa a e ANOVAs examining one ac o while holding he o he
cons an (see Supplemen a y) did no e eal a clea ly iden i iable
sou ce o he in e ac ions. These esul s sugges ha ou ask
design was e ec i e in inducing a ying le el o app oach-a oidance
con lic , he in e ac ion e ec s u he sugges ha pa icipan s
mus ha e conside ed bo h ewa d and punishmen p obabili ies.
A he in e -indi idual le el, we obse ed ha esponse ajec o ies
became mo e complex and a ying as app oach and a oidance
mo i a ions equals, such as in condi ions like .5 –.5 o .75 – .75.
Howe e , we did no obse e a clea -cu pa e n in mean beha iou al
indices ha could indica e uni o m condi ions o maximum and
minimum AAC ac oss pa icipan s.
EEG esul s
Time equency analysis
Wi h ou i s app oach o uni o m condi ions o maximum and
minimum AAC, we did no ind a ime- equency egion o signi i-
can condi ion con as in pe mu a ion and clus e es ing (Fig. 4).
Fu he mo e, when we compa ed MFT powe (4–8Hz) be ween uni-
o m maximum and minimum con lic condi ions in a ime win-
dow o 500–1100 ms, we did no ind a signi ican di e ence (see
Table 1 o s a is ics).
Howe e , he indi idualized app oach o con lic de ini ion
based on BCI e ealed signi ican ly s onge MFT powe in he max-
imum con lic condi ions compa ed o he minimum con lic
condi ions, (39) = 5.60, p < .001, d = 0.92 (Fig. 5, Table 1). The opo-
g aphy o his he a e ec was cen ed a mid- on al (FCz) si e.
Simila ly, indi idualized app oach based on RT and ewa d
ejec ion a e also yielded signi ican di e ence be ween MFT o
maximum and minimum con lic condi ion (Table 1). Howe e ,
he e was no signi ican egion a e pe mu a ion-clus e co ec-
ion in a e based app oach.
Sou ce sepa a ion
We applied a GED ha ex ac ed 65 componen s pe pa icipan .
Based on pe mu a ion es ing, only hose componen s exceeding
Figu e 3. The eac ion ime, he ewa d ejec ion a e, and he BCI as a unc ion o ewa d and punishmen p obabili y.
Figu e 4. Time- equency plo o he con as o uni o m maximum (0.5-0.5) and minimum AAC (1-0.25) condi ion and a opog aphical map o he a
(4–8 Hz) powe di e ences o his con as in he 500–1100 ms ime-window.
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026

6 | Pandey and Osinsky
he signi icance h eshold (α = 0.05) we e e ained (M = 3.13,
SD = 1.96). Nex , only componen s showing mid on al opog aphies
we e e ained. This p ocedu e e ealed mo e han one GED com-
ponen s o 60% o he pa icipan s, and no signi ican componen
o 5% o he pa icipan s (M = 1.75, SD = 0.89). The ime- equency
ac i i y o hese ex ac ed componen s shows an inc ease in he a
ac i i y a FCz (Fig. 6). We ound s ong co ela ions be ween di -
e ence in MFT powe o maximum and minimum con lic ials
o eal da a based on BCI and di e en ial MFT om ex ac ed GED
sou ces, = 0.68, p < .001. We did no ind a signi ican co ela ion
be ween di e en ial MFT om uni o m app oach and di e en ial
MFT om GED (Table 1). These esul s highligh ha he inc ease
in he a ac i i y in indi idually de ined maximum con lic condi-
ion compa ed o minimum con lic condi ion has mid- on al o igin.
We ound a mode a e co ela ion be ween BIS sco e and di e -
en ial MFT based on BCI (Fig. 7), = 0.35, P = .037 sugges ing ha
people ha ing highe BIS sco e we e mo e esponsi e o AAC, as
e lec ed in hei mid- on al he a ac i i y. Howe e , we did no
ind such signi ican co ela ion wi h MFT when applying a uni o m
de ini ion o con lic (Table 1).
MFT powe acks subjec i e con lic
We also examined whe he mid- on al he a (MFT) powe con in-
uously acks subjec i e con lic . To his end, we compu ed
wi hin-pa icipan co ela ions be ween he BCI and MFT powe
ac oss he 16 condi ions (Fig. 8a). The co ela ions we e signi ican ly
posi i e, indica ing ha highe subjec i e con lic was associa ed
wi h s onge MFT powe . The mean Fishe z- ans o med co ela-
ion was = 0.358, which was signi ican ly g ea e han ze o, (39) =
6.63, p < .001 (one- ailed). To u he es his ela ionship, we i ed
a linea mixed-e ec s model p edic ing subjec i e con lic (BCI)
om MFT powe , including a andom in e cep o pa icipan s.
The model e ealed a signi ican posi i e e ec o MFT powe on
con lic , b = 0.92, SE = 0.12, 95% CI [0.69-1.15], (638) = 7.85, p < .001.
The andom in e cep a iance ac oss pa icipan s was 0.025
(SD = 0.16), and model i indices we e AIC = –290.0 and BIC = –272.2.
Toge he , hese esul s demons a e ha MFT eliably acks sub-
jec i e con lic ac oss condi ions and pa icipan s.
In a nex s ep, we le e aged pa icipan s’ subjec i e con lic
expe iences o o de all 16 condi ions om highes o lowes
con lic . Fo each pa icipan , condi ions we e so ed based on hei
indi idual BCI sco es, such ha he i s condi ion ep esen ed
maximum con lic and he six een h condi ion ep esen ed mini-
mum con lic . This p ocedu e allowed us o a e age MFT powe
ac oss pa icipan s while p ese ing indi idual di e ences in con-
lic expe ience (see Supplemen a y o condi ion o de o each
pa icipan ). The esul ing analysis e ealed a g adual dec ease in
MFT powe om maximum o minimum con lic (Fig. 8b), p o iding
u he e idence ha MFT sys ema ically acks subjec i e con lic .
Discussion
This s udy examined neu al mechanisms o AAC p ocessing by
analysing MFT ac i i y ac oss a ying combina ions o ewa d and
punishmen p obabili ies. Mos impo an ly, we obse ed inc eased
MFT powe in high- compa ed o low-AAC condi ions, bu only i
condi ions we e de ined indi idually based on beha iou al indices
o con lic . We also ind ha MFT powe acks subjec i e con lic .
Compu a ional modelling (see supplemen a y) indica ed ha pa -
icipan s in eg a e bo h ewa d and punishmen in o ma ion in
decision-making, bu pa icipan s weigh hese ac o s di e en ly.
Mid- on al he a and con lic moni o ing
The obse ed inc ease in MFT powe du ing he high-AAC condi ion
aligns wi h exis ing e idence linking on al he a ac i i y o con lic
moni o ing and cogni i e con ol p ocesses. MFT ac i i y o he
pMFC has been iden i ied as a neu al ma ke o con lic de ec ion
and he ec ui men o con ol esou ces (Ca anagh and F ank 2014,
Lange e al. 2023). In he con ex o ou app oach-a oidance ask,
he indi idualized high-AAC condi ions likely elici ed heigh ened
compe i ion be ween app oach and a oidance d i es, necessi a ing
inc eased moni o ing and engagemen o con ol mechanisms. This
suppo s he no ion ha MFT ac i i y e lec s he dynamic egula-
ion o beha iou unde condi ions o unce ain y o compe ing
goals (e.g. Nigbu e al. 2011). Fu he mo e, in ou s udy MFT e ec s
pe sis o a a he leng hy window (up o 1200 ms) a e s imulus
onse , suppo ing he no ion ha hese oscilla ions no only e lec
con lic de ec ion bu also con inuous delibe a ion and esolu ion
be ween esponse op ions (Le y e al. 2023).
Indi idual di e ence in app oach a oidance
sensi i i ies
Ou esul s s ongly indica e ha indi idual di e ences in
app oach-a oidance sensi i i ies play a c ucial ole in con lic
p ocessing, highligh ing he impo ance o indi idualized a he
han p ede ining uni o m condi ions. Ou indings demons a e
ha when AAC was de ined uni o mly, MFT ac i i y did no
di e en ia e be ween maximum and minimum con lic ials
Table 1. MFT om maximum and minimum con lic condi ions based on Uni o m and Indi idualized app oaches o compu ing AAC,
and co esponding s a is ics.
Analysis ype
MFT (dB)
maximum
con lic
MFT (dB)
minimum
con lic
Whe he signi ian
egion a e
pe mu a ion-clus e
co ec ion
Compa ison o
maximum s.
minimum con lic
Co ela ion o
di e en ial
powe wi h
co esponding
GED compo-
nen s
ime se ies
Co ela ion o
di e en ial
powe wi h
BIS sco e
MSD MSD (39) p d p p
Uni o m
app oach
1–0.25 min 0.32 0.92 0.32 0.92 No 0.05 0.96 0.01 0.35 0.02 0.23 0.15
0.25–1 min 0.32 0.92 0.60 1.03 No 0.06 0.94 0.01 0.30 0.06 0.23 0.15
Indi idual-
ized
app oach
Ra e based 0.81 1.01 0.32 0.78 No 4.42 <.001 0.70 0.61 <.001 0.35 0.02
RT based 0.92 0.97 0.27 0.82 Yes 6.43 <.001 1.01 0.62 <.001 0.20 0.20
BCI based 0.84 1.00 0.28 0.81 Yes 5.48 <.001 0.86 0.67 <.001 0.35 0.02
We conside ed 0.5–0.5 condi ion as uni o m condi ion o maximum con lic while wo condi ions, 1–0.25 and 0.25–1, as uni o m condi ion o minimum con lic .
He e, i s alue ep esen s p obabili y o ewa d while second alue ep esen s p obabili y o punishmen .
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026
App oach-a oidance con lic | 7
ac oss pa icipan s. This sugges s ha p ede ined condi ions
may no accu a ely cap u e pa icipan s’ ac ual expe ience o
con lic , which is also suppo ed by ou beha iou al da a (i.e.
decision ime and choice a es). Ins ead, when con lic was
de ined indi idually based on eac ion ime and ewa d ejec ion
a e, MFT powe inc eased du ing high-con lic ials, indica ing
g ea e cogni i e con ol demands. This shi unde sco es he
idea ha indi iduals weigh ewa d and punishmen in o ma-
ion di e en ly in decision-making, in eg a ing bo h ac o s bu
assigning unique impo ance o each. Fo example, some pa -
icipan s may pe cei e a 50% - 75% ewa d-punishmen pai ing
as mo e con lic ing han a 75% o 50% pai ing, depending on
hei indi idual isk ole ance, decision s a egies, o mo e un-
damen al pe sonali y ac o s.
Ou esul s add o li e a u e ha sugges s ha MFT signals o
he pMFC no only e lec he need o con ol in “cold” cogni i e
p ocesses (like he p ocessing o simple s imulus esponse con lic )
bu may also pa ake in he esolu ion o mo e complex mo i a-
ional inconsis encies, which a e subjec o s ong indi idual di -
e ences and can be heo e ically amed by he Rein o cemen
Sensi i i y Theo y o pe sonali y (RST; G ay and McNaugh on 2000,
Co 2004). Acco ding o he RST, beha iou in a gi en si ua ion is
mainly d i en by he in e play o h ee independen sys ems o a
concep ual ne ous sys ems. The BAS is sensi i e o appe i i e cues
and ac i a es app oach beha iou . The FFFS is sensi i e o a e si e
cues and ac i a es a oidance beha iou . The BIS ecei es i s inpu
om bo h he FFFS and BAS is ac i a ed by mo i a ional con lic s
in gene al and app oach-a oidance con lic s in pa icula . I s main
Figu e 5. (a) Di e ence be ween he a ac i i y o maximum and minimum con lic condi ions based on indi idualized con lic de ini ion acco ding o
BCI sco es, a e aged ac oss pa icipan s, and co esponding opog aphical map a 4–8 Hz, 500–1100 ms. 0 ms e e s o he onse o he ba s-s imulus.
The do ed line ep esen signi ican egion a e pe mu a ion es ing and clus e co ec ion. The ec angle ep esen he window used o compu e
opog aphical plo . (b) The opog aphical plo o co esponding ime window. (c) The mid- on al he a ac i i y change o e ime. (d) The dis ibu ion o
change in he a ac i i y o indi idual pa icipan s.
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026
8 | Pandey and Osinsky
unc ion is o inhibi any ongoing beha iou and esol e con lic
by inc easing igilance and in o ma ion-seeking beha iou .
Ou indings sugges ha MFT signals o he pMFC may ack
subjec i e alence/app aisal o he ewa d and punishmen , and
hus be an impo an unc ional componen o he BIS. Fi s o all,
acco ding o he RST, he le el o con lic de ec ed by he BIS is
mainly d i en by an in e ac ion o inpu pa ame e s o he BAS
and FFFS (e.g. he objec i e p obabili ies o po en ial ewa ds and
punishmen s in a gi en si ua ion) and he gene al sensi i i ies o
hese wo ups eam sys ems. When inpu s om BAS and FFFS o
he BIS sys em a e iden ical, an indi idual will expe ience AAC only
when ewa d and punishmen a e equally likely. In con as , i
inpu om one o he wo sys ems o BIS is dominan , i would
need a highe p obabili y o he espec i e ein o cemen /punish-
men o ha e an iden ical FFFS and BAS inpu s o he BIS. The ac
ha we did no obse e a con lic e ec on MFT ampli ude ac oss
pa icipan s when compa ing uni o mly p ede ined condi ions bu
only when de ining condi ions on an indi idual basis, pe ec ly
aligns wi h his assump ion o BIS ac i a ion in he RST. Mo eo e ,
MFT could be a unc ional componen o BIS as we ound a signi -
ican co ela ion be ween sel - epo ed BIS sco es in he RST-PQ
and MFT. No ably many o he RST-PQ BIS i ems e e o s able
endencies o expe iencing anxious app ehension and/o a ousal.
Ou s udy is he e o e consis en wi h p io wo ks epo ing a
ela ion be ween MFT and indi idual di e ences in ai anxie y
(e.g. Ca anagh and Shackman 2015, Osinsky e al. 2017, Schmid
e al. 2018). T ansien inc eases in MFT a e also ela ed o he inhi-
bi ion o au oma ic app oach and a oidance beha iou (Ca anagh
e al. 2013, Swa e al. 2018). The BIS also modula e his au oma ic
app oach and a oidance beha iou acco ding o RST. Finally, he
BIS is hough o bias beha iou owa d a oidance, in case o a
non- esol able app oach-a oidance con lic , and s udies ha e
linked MFT esponse du ing AAC o his a oidance-biasing unc ion
o he BIS (Schmid e al. 2018, Ziebell e al. 2023).
Finding indi idualized con lic condi ions
In ou s udy, we employed h ee dis inc indi idualized app oaches
o iden i y ials o maximum and minimum AAC: one based on
ewa d ejec ion a e, one based on eac ion ime (RT), and hi d- a
composi e me ic BCI ha in eg a es bo h RT and ejec ion a e.
Ou esul s indica e ha he RT-based and BCI-based app oaches
a e pa icula ly e ec i e in cap u ing indi idualized AAC.
Howe e , we p opose ha an in eg a ed me ic like BCI can p o-
ides a mo e obus measu e o indi idualized AAC by accoun ing
o bo h decisional hesi a ion and ou come a iabili y. Impo an ly,
ou analysis showed ha o some pa icipan s, he highes RTs
occu ed in condi ions ha did no align wi h hei highes
beha iou al unce ain y (i.e. condi ions wi h ewa d accep ance/
ejec ion a es close o 50%). Con e sely, in o he cases, he lowes
RTs we e obse ed in condi ions ha we e no associa ed wi h
minimal unce ain y (i.e. ewa d ejec ion a es close o 0% o
100%). This dissocia ion unde sco es a c i ical limi a ion o elying
solely on ei he RT o ejec ion a e as a p oxy o con lic . While
RT cap u es he cogni i e demand associa ed wi h con lic , ewa d
ejec ion a e e lec s he beha iou al mani es a ion o decisional
unce ain y.
Ou ecommenda ion o BCI based app oach is consis en wi h
p io esea ch ha emphasizes he alue o mul idimensional
beha iou al indices o cap u ing la en cogni i e p ocesses. Fo
example, s udies ha e shown ha combining RT and choice
beha iou enhances he sensi i i y o compu a ional and
beha iou al models in decision-making con ex s (P e au e al. 2009,
Balla d and McClu e 2019, Zo owi z e al. 2019). In ligh o hese
indings, we ad oca e o he adop ion o composi e indices, such
as he BCI, in u u e s udies o app oach-a oidance con lic , pa -
icula ly when he goal is o cap u e indi idual a iabili y in con lic
p ocessing wi h g ea e ideli y.
Fu u e di ec ions
In ou esul s we in e p e he s imulus-locked inc ease in mid- on al
he a as an index o con lic moni o ing, howe e , al e na i e mech-
anisms may also ha e con ibu ed. Fo ins ance, mid- on al he a
has been linked o unce ain y and expec ancy iola ion, aising he
possibili y ha he obse ed e ec may e lec s pa icipan s’
unce ain y abou he ial ou come a he han con lic . Ano he
explana ion is ha indi idual di e ences in he subjec i e alence/
salience o ce ain classes o ewa ds and punishmen s could d i e
he a a iabili y. Since we did no di ec ly measu e subjec i e
salience/ alence (e.g. by sel - epo ed a ings o punishmen and
ewa d s imuli) and only used one ype o ewa d (poin s con e ible
o money) and one ype o punishmen (unpleasan sounds), ou
s udy is no sui ed o add ess his issue. Punishmen expec ancy is
Figu e 6. Componen ime- equency ac i i y o GED mid on al
componen s a e aged ac oss subjec s based on BCI.
Figu e 7. Co ela ion be ween BIS sco e and di e en ial MFT
based on BCI.
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026
App oach-a oidance con lic | 9
ano he p ocess ha could also lead o enhanced he a powe inde-
penden o con lic . Fu he mo e, i is possible ha p olonged
esponse imes, e lec ing slowe e idence accumula ion, pa ially
accoun o he e ec . Finally, he a inc eases migh index an icipa-
o y con ol p ocesses associa ed wi h p epa ing o a oid nega i e
ou comes, a he han he simul aneous co-ac i a ion o compe ing
esponses. Al hough in ou supplemen a y analysis, we show ha
ou esul s a e no d i en by an icipa ion o punishmen o longe
eac ion ime, hese al e na i e pe spec i es unde sco e he need o
u u e wo k ha manipula es unce ain y, ou come alence, and
empo al dynamics mo e di ec ly o disen angle hei con ibu ions
o mid- on al he a. A g ea e unde s anding o AAC can yield
impo an insigh s o p ac ical ele ance in he ield o maladap i e
a oidance, he eby helping o build a b idge be ween expe imen al
and applied esea ch (Ga cia-Gue e o e al. 2023). Fu u e esea ch
may u he explo e how pe sonali y ai s in luence indi idual sen-
si i i ies o app oach, a oidance, he con lic he eo and co espond-
ing neu al esponses. Fu u e s udies should inco po a e indi idual
measu es a he han elying solely on p ede ined con lic condi ions.
Mo eo e , u he esea ch may u he explo e connec i i y be ween
mid on al and pa ie al egions o unde s and how he a and
be a- ela ed neu al mechanisms in e ac in con lic esolu ion.
Con lic o in e es : The au ho s decla e ha hey ha e no compe -
ing in e es s.
Funding
This esea ch was no unded by any speci ic unding.
Da a a ailabili y
The da a, da a summa y, and code used a e a ailable h ough OSF:
h ps://os .io/95 8j/o e iew? iew_only= a5062a33d4e4ab7a4488
ab3d 1ca050.
Au ho con ibu ions
Shubham Pandey (Concep ualiza ion, Fo mal analysis, In es iga-
ion, Me hodology, W i ing—o iginal d a , W i ing— e iew & edi -
ing, Roman Osinsky (Concep ualiza ion, Funding acquisi ion,
Me hodology, Resou ces, Supe ision, W i ing— e iew & edi ing)
Acknowledgemen s
We hank ou s uden s assis an s, Tim Bee mann, Julia Wo mann,
and Joshua Thie , o help wi h da a collec ion.
Re e ences
Ahn WY, K awi z A, Kim W e al. A Model-Based MRI analysis wi h
hie a chical Bayesian pa ame e es ima ion. J Neu osci Psychol Econ
2011;4:95–110.
Balla d IC, McClu e SM. Join modeling o eac ion imes and choice
imp o es pa ame e iden i iabili y in ein o cemen lea ning mod-
els. J Neu osci Me hods 2019;317:37–44.
Ca anagh JF, Eisenbe g I, Gui a -Masip M e al. F on al he a o e ides
Pa lo ian lea ning biases. J Neu osci 2013;33:8541–8.
Ca anagh JF, F ank MJ. F on al he a as a mechanism o cogni i e
con ol. T ends Cogn Sci 2014;18:414–21.
Ca anagh JF, Cohen MX, Allen JJB. P elude o and esolu ion o an e o :
EEG phase synch ony e eals cogni i e con ol dynamics du ing
ac ion moni o ing. J Neu osci 2009;29:98–105.
Ca anagh JF, Shackman AJ. F on al midline he a e lec s anxie y and
cogni i e con ol: me a-analy ic e idence. J Physiol Pa is 2015;
109:3–15.
Cohen MX. A u o ial on gene alized eigendecomposi ion o denois-
ing, con as enhancemen , and dimension educ ion in mul i-
channel elec ophysiology. Neu oimage 2022;247:118809.
Cohen MX, Donne TH. Mid on al con lic - ela ed he a-band powe
e lec s neu al oscilla ions ha p edic beha io . J Neu ophysiol
2013;110:2752–63.
Co PJ. Rein o cemen sensi i i y heo y and pe sonali y. Neu osci
Biobeha Re 2004;28:317–32.
Co PJ, Coope AJ. The ein o cemen sensi i i y heo y o pe sonali y
ques ionnai e (RST-PQ): de elopmen and alida ion. Psychol Assess
2016;28:1427–40.
Dup ez J, Gulbinai e R, Cohen MX. Mid on al he a phase coo dina es
beha io ally ele an b ain compu a ions du ing cogni i e con ol.
Neu oimage 2020;207:116340.
Ga cia-Gue e o S, O’Ho a D, Zgonniko A e al. The ac ion dynamics
o app oach-a oidance con lic du ing decision-making. Q J Exp
Psychol (Ho e) 2023;76:160–79.
G ay JA, McNaugh on N. (2000). The Neu opsychology o Anxie y: An
Enqui y in o he Func ions o he Sep o-Hippocampal Sys em (2nd ed.).
Ox o d Uni e si y P ess.
Figu e 8. MFT powe acks subjec i e con lic . (a) Wi hin-pa icipan s co ela ion o MFT wi h con lic sco e (BCI); (b) G adual dec ease o mid- on al
he a (MFT) powe om maximum o minimum indi idualized con lic .
Downloaded om h ps://academic.oup.com/scan/a icle/20/1/nsa 114/8307139 by Uni e si a Osnab uck use on 19 Feb ua y 2026