Neu obiology o Lea ning and Memo y 183 (2021) 107476
A ailable online 2 June 2021
1074-7427/© 2021 The Au ho s. Published by Else ie Inc. This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/).
Wo king memo y load modula es oscilla o y ac i i y and he dis ibu ion o
as equencies ac oss on al he a phase du ing wo king
memo y main enance
Alba Fe n´
andez
a
,
*
,
1
, Diego Pinal
b
,
1
, Fe nando Díaz
a
, Mon se a Zu ´
on
a
a
Cogni i e Neu oscience Labo a o y, Depa men o Clinical Psychology and Psychobiology, Uni e sidade de San iago de Compos ela, San iago de Compos ela, Galiza,
Spain
b
Psychological Neu oscience Lab, Escola de psicologia, Uni e sidade do Minho, Po ugal
ARTICLE INFO
Keywo ds:
Neu al oscilla ions
Wo king memo y
C oss- equency synch oniza ion
ABSTRACT
Wo king memo y (WM) is a keys one o ou cogni i e abili ies. Inc easing load has been shown o dampen i s
pe o mance and a ec oscilla o y neu al ac i i y in di e en equency bands. Ne e heless, mixed esul s
ega ding as equencies ac i i y and a lack o esea ch on WM load modula ions o c oss- equency phase-
ampli ude coupling mechanisms p eclude a be e unde s anding o he impac o inc eased WM load le els on
b ain ac i i y as well as in e - egional communica ion and coo dina ion suppo ing WM p ocesses. Hence, we
analyzed he EEG ac i i y o 25 pa icipan s while pe o ming a delayed-ma ching- o-sample (DMS) WM ask
wi h h ee WM load le els. Cu en densi y powe and dis ibu ion a he sou ce le el o he a, be a, and gamma
equencies du ing he ask’s delay pe iod we e compa ed o each pai o WM load condi ions. Resul s showed
maximal inc eases o he a ac i i y in on al a eas and o as equencies’ ac i i y in pos e io egions wi h WM
load, showing he in ol emen o on al he a ac i i y in WM main enance and he con ol o a en ional e-
sou ces and isual p ocessing by be a and gamma ac i i y. To s udy whe he WM load modula es communica ion
be ween co ical a eas, pos e io be a and gamma ampli udes dis ibu ion ac oss on al he a phase was also
analysed o hose a eas showing he la ges signi ican WM load modula ions. Highe be a ac i i y ampli ude a
bila e al cuneus and igh middle occipi al gy us, and highe gamma ac i i y ampli ude a bila e al pos e io
cingula e we e obse ed du ing on al he a phase peak in low han high memo y load condi ions. Mo eo e ,
g ea e as be a ampli ude a he igh pos cen al gy us was obse ed du ing he a phase ough a igh middle
on al gy us in high han low memo y load condi ions. These esul s show ha WM load modula es whe he
in e egional communica ion occu s du ing heo e ically op imal o non-op imal ime windows, depending on
he demands o on al con ol o pos e io a eas equi ed o pe o m he ask success ully.
1. In oduc ion
Wo king memo y (WM) is he capaci y ha allows us o e ain and
manipula e o b ie pe iods o ime small amoun s o in o ma ion no
longe a ailable in he en i onmen (Baddeley, 1998, 2003). WM in-
ol es di e en subp ocesses, namely, he ini ial encoding o in o ma-
ion in empo a y s o es, he main enance du ing a ew seconds o i s
ep esen a ion in hose s o es and, inally, he e ie al o his in o -
ma ion o i s use.
Di e en esea ch has s udied WM using
delayed–ma ching– o–sample (DMS), n-back, and S e nbe g asks.
Among hese, one o he mos used asks o s udy WM subp ocesses is he
DMS ask, in which pa icipan s mus memo ize non- e balizable pa -
e ns o i ems and, a e a b ie delay, co ec ly ma ch hem o a p e-
sen ed p obe pa e n. This ask can also be used o s udy WM load’s
e ec by inc easing he numbe o i ems o be memo ized. Such in-
c eases in WM load a e accompanied by dec eased ask pe o mance,
e lec ed in longe eac ion imes and educed esponse accu acy a es
(Al a ez & Ca anagh, 2004; Vogel, Woodman, & Luck, 2001).
Elec oencephalog aphy (EEG) is a echnique ha allows eco ding
neu oelec ic oscilla o y ac i i y. Thus, i enables he explo a ion o he
neu al bases o WM subp ocesses, such as main enance, and he e ec s
* Co esponding au ho .
E-mail add ess: [email p o ec ed] (A. Fe n´
andez).
1
Con ibu ed equally o his wo k.
Con en s lis s a ailable a ScienceDi ec
Neu obiology o Lea ning and Memo y
jou nal homepage: www.else ie .com/loca e/ynlme
h ps://doi.o g/10.1016/j.nlm.2021.107476
Recei ed 2 No embe 2020; Recei ed in e ised o m 13 May 2021; Accep ed 30 May 2021
Neu obiology o Lea ning and Memo y 183 (2021) 107476
2
o WM load on hem. S udies ha employ EEG oscilla o y ac i i y ana-
lyses ha e demons a ed ha oscilla o y neu al ac i i y in slow e-
quency bands such as he a (4 – 7/8 Hz) may play a key ole in
empo a ily main aining in o ma ion in WM s o es. P obably he a ac-
i i y eco ded in on al elec odes e lec s b ain ac i i y ela ed o
a en ional con ol o e he s o ed in o ma ion o ano he kind o cen-
al execu i e unc ions ( o a e iew, see Sauseng, G iesmay , F eun-
be ge , & Klimesch, 2010).
Rega ding as oscilla ions (i.e., equencies o e 13 Hz), many
s udies ha e obse ed hei p esence h oughou a ious pos e io
co ical egions du ing he main enance pe iod in WM asks. Howe e ,
he esul s o hese s udies a e inconsis en as o how be a and gamma
ac i i y a e implica ed in WM p ocesses ( o a e iew see Pa lo &
Ko choubey, 2020). Hence, some s udies ha e epo ed inc eases in hei
ampli ude ela i e o a baseline pe iod in isual and occipi o empo al
a eas (Honkanen, Rouhinen, Wang, Pal a, & Pal a, 2015; Tallon-
Baud y, Be and, Pe onne , & Pe nie , 1998), while o he wo k e-
po s supp ession o hese equencies in hose same a eas du ing he
execu ion o simila asks (B ookes e al., 2012; P osko ec, Wiesman,
Hein ichs-G aham, & Wilson, 2018). These esul s, howe e , a e ob-
ained in pa adigms wi h a single WM load condi ion.
WM load inc eases he demands made on neu al esou ces o achie e
a success ul ask pe o mance. Thus, a ia ions in i s le el a e expec ed
o impac oscilla o y ac i i y ela ed o WM p ocessing. Indeed, he a
oscilla o y ac i i y has been shown o be modula ed by his ac o .
The e o e, he a ampli ude in a spa ial e sion o an n-back ask (in
which pa icipan s mus e ain a sequence o i ems o judge whe he he
cu en i em ma ches an i em n places back in he sequence) was la ge
a on al midline si es as WM load demands (i.e., he numbe o i ems
back) inc eased (Ge ins, 1997). The a ac i i y has also been ound o
inc ease wi h WM load o e on al midline si es du ing he delay pe iod
o a DMS ask (Eschmann, Bade , & Mecklinge , 2018) as well as o a
S e nbe g ask (pa icipan s a e asked o memo ize a lis o i ems o
decide upon p esen a ion o a p obe i i was included on he p e ious
lis ) wi h non e bal symbols (Mau e e al., 2015). Neu al sou ces o his
he a ac i i y ha e been iden i ied a he medial p e on al co ex
(Kaplan e al., 2016; an Ede, Jensen, & Ma is, 2017)
Howe e , wi h be a and gamma oscilla ions, o e all he pic u e is
no as clea ( o a e iew see Pa lo & Ko choubey, 2020). When
looking a he in luence o WM load in be a and gamma ac i i y du ing
he main enance pe iod o WM asks, he esul s epo ed in p e ious
s udies a e inconsis en , showing bo h inc eases and supp ession o his
ac i i y depending on he speci ic wo k. Some s udies ha e obse ed
inc eased be a and gamma ampli ude wi h WM load. Fo example, in a
s udy wi h in ac anial eco dings, inc eased be a ampli ude du ing he
main enance pe iod o a S e nbe g ask was ound o e p e on al, pa-
ie al, and empo al a eas o longe han sho e lis s o le e s p e-
sen ed sequen ially (Howa d e al., 2003). Simila ly, using a DMS ask
whe e pa icipan s had o memo ize a sample s imulus wi h 1 o 6
squa es, gamma ampli ude du ing he main enance pe iod was also
obse ed o inc ease wi h g ea e WM loads o e on al, pa ie al, and
empo al egions in MEG da a analyses (Pal a, Kulashekha , Hamalai-
nen, & Pal a, 2011).
On he o he hand, a simila numbe o s udies ha e ound ei he no
signi ican change in as oscilla o y ac i i y wi h WM load changes o
e en a dec ease o be a and/o gamma ac i i y wi h WM load modula-
ions. Fo example, P osko ec, Wiesman, Hein ichs-G aham, and Wilson
(2019) ound a educ ion o be a ac i i y wi h inc eased WM load in
pos e io egions du ing he main enance pe iod o a isual DMS ask
whe e pa icipan s had o memo ize he loca ion o 2 o 4 black squa es
loca ed in a 7 x 9 g id. Also, ega ding gamma oscilla ions, Poch, Campo,
and Ba nes (2014) ound no e ec o WM load on i s ampli ude du ing
he main enance pe iod o a isual e o-cueing WM ask whe e pa ic-
ipan s had o memo ize he loca ion and o ien a ion o 1, 2, o 4 ec -
angles o la e ma ch hem wi h a p obe. Likewise, Paho and Jauˇ
so ec
(2017) ound no e ec o WM load on gamma ac i i y du ing he
main enance pe iod o a S e nbe g ask whe e pa icipan s had o
memo ize a isual a ay o 4, 6, o 8 colo ed squa es.
Fas EEG equencies a e less well s udied han he a ac i i y, espe-
cially wi h non-in asi e echniques such as scalp eco ded EEG, which
may be one con ibu ing ac o o hese mixed esul s. Fu he , e en
when he same o simila asks a e used in di e en s udies, he e a e
s ill di e ences in he pa icula pa ame e s selec ed (i.e., he numbe o
and he ea u es o he i ems o e ain in memo y o each WM load
condi ion, he ype o s imuli used, he du a ion o he delay pe iod,
e c.). Gi en ha bo h equencies a e in ol ed in many cogni i e p o-
cesses beyond WM main enance, i is di icul o dissec wha is p ecisely
d i ing he inc ease o dec ease o his ac i i y. Addi ionally, only a ew
o he p e iously men ioned s udies ha e a emp ed o iden i y he
co ical o igins o such scalp- eco ded EEG oscilla o y ac i i y.
In summa y, se e al s udies epo ha WM load modula es hese
as equencies. Howe e , he di ec ion o his modula ion emains
unclea , and i seems o be mo e widely dis ibu ed in he scalp han in
he case o he a oscilla ions (Honkanen e al., 2015; Howa d e al.,
2003; Pal a e al., 2011; P osko ec e al., 2019).
Besides he independen esponse o he a, be a, and gamma bands o
WM p ocesses and hei WM load modula ion, he e has been a g owing
in e es in he in e ac ion be ween slow ( he a) and as oscilla o y (be a
and gamma) ac i i y du ing WM p ocessing. This in e es s ems om
ecen heo e ical models highligh ing he ele ance o he in e ac ion
be ween oscilla o y b ain ac i i y in di e en equencies o neu al
communica ion and ac i i y coo dina ion (e.g., Canol y & Knigh , 2010;
Roux & Uhlhaas, 2014). Expe imen al e idence in suppo o hese
models includes epo s o slow equencies phase and as oscilla o y
ac i i y phase o ampli ude synch oniza ion in n-back (Yang & Huang,
2018) and DMS asks as eco ded wi h EEG (Sauseng e al., 2009) as well
as wi h MEG (Siebenhühne , Wang, Pal a, & Pal a, 2016). Fo example,
i has been obse ed ha wi h highe memo y loads, he e is an inc ease
o he a – gamma phase synch oniza ion du ing WM main enance ha is
una ec ed by i ele an i ems and is posi i ely co ela ed wi h he in-
di iduals’ WM capaci y (Sauseng e al., 2009).
Rega ding he synch oniza ion p oduced be ween he phase o a slow
equency (e.g., he a) and he ampli ude o a as e equency (e.g.,
gamma o be a) in he same o a di e en co ical a ea; heo e ical
models o b ain unc ion sugges ha communica ion be ween neu onal
popula ions depends on hese g oups being phase synch onized and hus
ha ing hei windows o inpu and ou pu open simul aneously (F ies,
2005). The e o e, i is hypo hesized ha he slow equency cycle will
de e mine ime windows o enhanced o dec eased ecep i i y o
es ablish unc ional links (F ies, 2005). These unc ional links will be
es ablished whene e he pe iods o highe ampli ude o he as e
oscilla ion co-occu wi h a de e mined phase angle o he slowe e-
quency in he same o di e en b ain egions. In his ega d, i is
belie ed ha neu onal popula ions will be mo e o less likely o be
exci ed, depending on he phase angle o he slow oscilla o y ac i i y.
Speci ically, hey a e supposed o be mo e likely o be exci ed du ing he
ough and less likely du ing he peak (F ies, 2005). When his phe-
nomenon is p oduced be ween dis an b ain egions, i is known as
in e egional phase – ampli ude coupling (PAC). This phenomenon has
been p oposed o play an essen ial ole in memo y p ocesses in gene al
and in WM in pa icula (Roux & Uhlhaas, 2014).
Fu he mo e, ega ding WM load’s in luence in his mechanism,
in ac anial eco dings in he human hippocampus ha e shown ha he
speci ic he a equency a which gamma ampli ude was coupled
dec eased as a unc ion o WM load (Axmache e al., 2010). E idence
o his he a-gamma mechanism in memo y p ocesses, and in WM
speci ically, was p o ided by a s udy in which ansc anial al e na ing
cu en s imula ion ( ACS) was used o al e he indi idual a io o he a
o gamma equencies. To ha end, he au ho s induced slowe he a
hy hms han he indi iduals’ na u al he a equency showing
maximum coupling o gamma ampli udes in a subse o he expe imen al
pa icipan s. Those pa icipan s ecei ing ACS would hen had mo e
A. Fe n´
andez e al.
Neu obiology o Lea ning and Memo y 183 (2021) 107476
3
gamma cycles p oduced wi hin a he a cycle, which was hypo hesized o
inc ease hei WM span. Indeed, hose pa icipan s demons a ed an
inc eased digi span du ing s imula ion han be o e i , while he pe -
o mance o hose pa icipan s ecei ing sham s imula ion did no
change (Vosskuhl, Hus e , & He mann, 2015).
Ne e heless, he obse ed modula ion o equency wi hin he a
ange in PAC mechanisms does no p o ide any in o ma ion o e idence
ega ding he he a ac i i y phase angle a which as oscilla o y ac i i y
is coupled. Indeed, o he bes o ou knowledge, i has no been p e i-
ously s udied whe he WM load could modula e he he a equency
phase window a which as oscilla o y ac i i y (be a and gamma) is
nes ed. This would shed new insigh s on how co ical a eas communi-
ca e depending on WM load.
In his wo k, ou main goal is s udying whe he WM load in luences
in e egional co ical communica ion be ween he b ain egions
showing he la ges modula ion o oscilla o y ac i i y by WM load. To do
so, we will analyse how WM load modula es he ela ionship be ween
ampli udes o as equencies (be a and gamma) wi h he phase o a
slowe equency (i.e., he a) in o sou ce space.
Wi h his aim in mind, in he i s pa o he manusc ip we will use
sou ce es ima ion o in es iga e he neu al sou ces ha show he la ges
e ec s o WM load manipula ions in he he a, be a and gamma e-
quency bands du ing he main enance o in o ma ion in wo king
memo y. In he second pa o he manusc ip , we will s udy he co ical
communica ion be ween hese egions.
In o de o do his, oscilla o y b ain ac i i y was eco ded du ing he
main enance pe iod o a isual DMS ask wi h h ee di e en memo y
load le els, and sou ce es ima ion was used o ans o m scalp eco ded
ac i i iy in o sou ce space ac i i y. A DMS ask was selec ed because i
allowed us o accu a ely delimi he main enance pe iod in WM wi h no
in e e ing p ocesses.
1.1. The wo king hypo heses we e:
(1) WM load should modula e he a and as equencies (be a and
gamma) ampli udes du ing he main enance pe iod o a DMS
ask; and o he a, i was expec ed ha ampli ude would inc ease
as a unc ion o WM load.
(2) The dis ibu ion a he sou ce space le el o he la ges di e -
ences be ween WM load condi ions should di e among e-
quencies. Fo he a, i was expec ed ha he maximum inc ease in
powe wi h WM load would be loca ed in p e on al a eas.
(3) Communica ion be ween a eas wi h he la ges di e ence in
powe be ween WM load condi ions should also be modula ed by
WM load. I was expec ed ha WM load modula es he dis ibu-
ion o as equencies ampli ude ac oss he a phase, in a eas
whe e he modula ion o WM load was maximal. Speci ically: o
each pai o communica ing a eas, we expec ed ha be a and
gamma ampli udes du ing he a ough we e g ea e in highe
WM load condi ions compa ed o hei ampli udes in lowe WM
load condi ions. Consequen ially, i was also expec ed ha be a
and gamma ampli udes du ing he a peak, we e g ea e in lowe
WM load condi ions compa ed wi h highe WM load condi ions.
2. Me hods
2.1. Pa icipan s
A o al o hi y- wo heal hy olun ee s (20 women, mean age 21 ±
3.3 yea s old) pa icipa ed in he s udy. All o hem we e igh -handed as
assessed by he Edinbu gh Handedness In en o y (Old ield, 1971) and
had no mal o co ec ed- o-no mal ision and no his o y o psychia ic
o neu ological diso de s. Se en women had o be excluded om da a
analysis due o excessi e a e ac s in he EEG da a (see sec ion 2.4 o
de ails). The emaining 25 pa icipan s had a mean age o 21.1 ±2.70
yea s old.
Pa icipan s epo ed no ecen consump ion o alcohol o ec ea-
ional d ugs and we e ins uc ed o abs ain om ca eine and o he
s imulan s a leas 30 min be o e he EEG eco ding. All pa icipan s
signed an in o med consen o m be o e he expe imen al session. The
s udy was app o ed by he e hics on in es iga ion au onomic commi ee
o Galicia (CAEIG, Galicia, Spain; code: 82017/498).
2.2. Task
Pa icipan s pe o med a delayed-ma ching- o-sample isuospa ial
ask, p eceded by a 6- ials aining block in he same ask. The ask
con ained 120 ials di ided in o h ee blocks wi h a 90-second in e al
be ween blocks. Each ial began wi h a wa ning sound (1000 Hz, 150
ms du a ion). A e a ji e ed p e-s imulus in e al o 450–550 ms, a
sample s imulus (encoding s age) appea ed, and emained 1750 ms on
sc een. The sample s imulus consis ed o h ee domino iles, and pa -
icipan s we e ins uc ed o memo ize he numbe o do s and hei
posi ion wi hin each ile. A e a main enance pe iod (3250–3750 ms), a
p obe s imulus (in o ma ion e ie al s age) appea ed on sc een o
1750 ms. The p obe s imulus also comp ised h ee domino iles ha
could ma ch he sample s imulus o ha e one o he do s in a di e en
loca ion wi hin one o he domino iles. Pa icipan s had o decide by
bu on-p ess whe he he p obe s imulus ma ched he p e iously enco-
ded sample s imulus o no . Be ween ials, he e was a 900–1100 ms
ji e ed in e al (Fig. 1).
Each ile was an 8 ×4 cm ec angle o med by wo whi e squa es,
p esen ed on he cen e o a la -sc een loca ed a a dis ance o 1 m om
he pa icipan , co e ing a isual angle o 4.58◦x 2.29◦. Tiles could ha e
be ween 3 and 4 do s andomly dis ibu ed ac oss he ou co ne s o
each squa e (i.e., eigh di e en possible loca ions). The do s we e black
and we e loca ed 0.5 cm om he ile edges and 1 cm om each o he .
Fu he mo e, he numbe o do ed iles in he sample and p obe s imuli
was manipula ed o c ea e h ee di e en wo king memo y load con-
di ions. Gi en ha only EEG epochs wi h co ec answe s would be
analysed and ha a la ge numbe o e o s was expec ed in he highe
load condi ions, he numbe o ials in each condi ion was adjus ed.
Hence, mo e ials we e included o condi ions wi h highe loads o
compensa e o hese ac s and s ill e ain a simila numbe o alid
epochs in each condi ion o pos e io analysis. Thus, on he low
wo king memo y load (LL) condi ion, bo h iles a he sides we e blank
-i.e., emp y o do s- (30 ials); on he medium WM load (ML) condi ion,
he cen e ile was blank (40 ials), and on he high WM load (HL)
condi ion all iles had do s inside (50 ials). S imulus’ p esen a ion was
semi- andomized so ha a maximum o h ee “ma ch” o “non-ma ch”
consecu i e ials could appea . 50% o ials we e “ma ch”, and 50%
we e “non-ma ch” ials.
2.3. Beha iou al da a
Pe cen age o co ec esponses (hi s) and eac ion imes measu ed
om p obe s imulus onse o bu on p ess only om ials wi h co ec
esponses we e calcula ed o each pa icipan and WM load condi ion.
2.4. EEG eco ding and signal p ocessing
Pa icipan s sa in a com o able a mchai inside a noise and ligh
a enua ed Fa aday chambe . EEG ac i i y was eco ded be ween 0.001
and 100 Hz wi h a 50 Hz no ch il e and digi ized a 500 Hz om 60 Ag-
AgCl ac i e scalp elec odes posi ioned acco ding o he 10–10 sys em,
wi h nose ip e e ence and he g ound elec ode a Fp1 loca ion. Be-
sides, e ical and ho izon al EOG we e eco ded wi h wo elec odes
posi ioned on he ou e can hi o bo h eyes (HEOG); and wo elec odes
placed abo e and below he igh eye (VEOG). Elec ode impedances
we e kep below 10 kΩ.
All 32 con inuous da a eco dings we e inspec ed isually and wi h
he Raw Da a Inspec ion u ili y in B ain Vision Analyze o de e mine
A. Fe n´
andez e al.
Neu obiology o Lea ning and Memo y 183 (2021) 107476
4
hei sui abili y o u he analysis. The ollowing pa ame e s we e use
in Raw Da a Inspec ion: o all 60 scalp elec odes, i would ma k as bad
sec ions o da a om 200 ms be o e o 200 ms a e a di e ence la ge
han ±200
μ
V in any 200 ms window. I would also ma k ol age s eps
la ge han 50
μ
V and segmen s wi h 0.5
μ
V o lowe ampli udes o 100
ms o longe in e als. Reco dings ma ked wi h one o mo e o hese
a e ac s in mo e han 20% o he en i e da a o mo e han 6 channels
we e excluded om u he analysis. As a esul , eco dings o 7 emale
pa icipan s we e disca daded om u he analyses lea ing a sample
size o 25 pa icipan s.
Da a om he emaining 25 pa icipan s we e hen o line esampled
o 512 Hz o op imize hem o pos e io ime– equency analysis. A
phase shi - ee Bu e wo h il e be ween 0.5 and 80 Hz (12 dB/oc a e
oll-o ) was applied. Ocula a e ac s we e co ec ed using independen
componen analysis (ICA) in B ain Vision Analyze , and only compo-
nen s compa ible wi h ocula mo emen s we e ejec ed. Du ing his
s ep, all componen s and opog aphic maps o each pa icipan we e
also e ised in o de o con ol and e i y ha componen s picking up
low ampli ude muscula noise (i.e., ex acephalic componen s wi h
equency peaks abo e 30 Hz) o o he noise sou ces (e.g. noisy chan-
nels) we e no en e ed in signal econs uc ion o u he analyses.
EEG da a we e segmen ed in 5500 ms epochs om sample s imulus
onse (i.e., comp ising he main enance pe iod) o hose ials wi h
co ec answe s. Semi-au oma ic a e ac ejec ion was also applied o
exclude EEG segmen s wi h da a poin s exceeding ±125 µV. Epochs
we e u he g ouped in o low, medium, and high wo king memo y load
(LL, ML, and HL) ials o each pa icipan (mean epochs o LL =27.64
±2.14; mean epochs o ML =32.44 ±3.20; mean epochs o HL =
32.84 ±4.38).
3. Da a analyses and esul s
Due o pa o he analyses being dependen on he esul s o p e ious
s eps, i was decided o g oup bo h analysis and esul s in he same
sec ion in o de o acili a e hei eading.
3.1. Beha iou al analyses and esul s
To es o po en ial e ec s o WM load on ask pe o mance, pe -
cen age o hi s and eac ion imes we e subjec ed o epea ed-measu es
ANOVAs wi h he wi hin-subjec s ac o WM Load (LL, ML, HL).
G eenhouse–Geisse co ec ion was used when he sphe ici y assump-
ion was iola ed, and Bon e oni co ec ion was employed o pos hoc
pai wise compa isons.
Visual wo king memo y load signi ican ly dampens ask pe o -
mance, o bo h pe cen age o hi s F(2,48) =134; p <0.001, and e-
ac ion imes F(1.5, 36.3) =139.8; p =0.01. Pa icipan s’ eac ion imes
we e signi ican ly longe o HL (mean =1115.6 ms, SD =234.5 ms)
han bo h, ML (mean =980.7 ms, SD =210 ms) and LL (mean =709.4
ms, SD =130.2 ms) condi ions (p <0.001 and p <0.001, espec i ely);
as well as o ML han LL (p <0.001). In addi ion, pe cen age o hi s
signi ican ly dec eased wi h highe WM load (LL: mean =95.1%, SD =
5.5; ML: mean =84.1%, SD =8.8; HL: mean =67%, SD =9). Thus,
pe cen age o hi s was signi ican ly lowe o HL han ML and LL (p <
0.001 and p <0.001, espec i ely) and lowe o ML han LL (p <0.001).
3.2. Cu en densi y powe analyses and esul s
The ini ial epoch was u he limi ed o a 2000 ms epoch ollowing
he o se o he sample s imulus, so ha only he main enance pe iod
was included. In o de o assess po en ial WM load- ela ed di e ences
on oscilla o y ac i i y powe a he co ical sou ces o scalp- eco ded
EEG signals, s anda dized low- esolu ion elec omagne ic omog aphy
analysis (sLORETA) so wa e (Pascual-Ma qui, 2002) was used.
Fi s , o each single equency and elec ode pai in he epoched
da a, sLORETA uses he mean ol age alues o calcula e he c oss-
spec a, which is hen a e aged ac oss epochs. Thus, a ull c oss-
spec a ma ix is ob ained o each pa icipan . These c oss-spec a
ma ices a e eed o sLORETA algo i hms o calcula e he cu en
sou ce densi y powe es ima es o 6430 oxels o 5 ×5 mm ep e-
sen ing co ical g ey ma e and hippocampus, based on a 3-shell
sphe ical head model egis e ed o he Talai ach b ain a las (Pascual-
Ma qui, 2002).
Finally, di e ences in mean cu en densi y powe be ween each pai
o WM load condi ions (i.e. HL s LL, HL s ML, and ML s LL) we e
e alua ed o each o he ollowing equency bands: he a (4–7 Hz),
be a 1 (13 – 20 Hz), be a 2 (20–30 Hz), gamma 1 (30–50 Hz) and gamma
2 (50–70 Hz) using s a is ical non-pa ame ic mapping (SnPM) as
implemen ed in sLORETA. Hence, independen analyses we e un o
compa e cu en densi y powe maps o each pai o WM condi ions and
equency band (i.e., 15 independen SnPM analyses) using 5000 an-
domiza ions o pai ed samples - es s on log ans o med da a. Each one
o hese andomiza ions calcula ed 6430 pai ed samples - es s (one o
each co ical g ay ma e oxel) and s o es he wo mos ex eme alues
(i.e., one pe ail o he dis ibu ion) o he -s a is ics o build a dis i-
bu ion o ex eme -s a is ics. The -s a is ic ob ained wi h he obse ed
Fig. 1. DMS ask used in he expe imen .
A. Fe n´
andez e al.
Neu obiology o Lea ning and Memo y 183 (2021) 107476
5
da a was hen compa ed agains his dis ibu ion o ex eme s a is ics o
assess i s s a is ical signi icance using an alpha h eshold se a p <0.05.
This p ocedu e allows us o co ec o mul iple compa isons when
es ablishing he signi icance o he obse ed sco e and do no equi e
he assump ion o Gaussiani y ( o de ails see Nichols & Holmes, 2002).
All he gy i con aining signi ican oxels in he B odmann a ea (BA)
showing he mos ex eme obse ed -s a is ic o each SnPM es (see
Table 1 and Fig. 2) we e selec ed as egions o in e es (ROIs) o he
in e egional c oss- equency coupling analyses. A comp ehensi e lis o
he esul s o hese cu en densi y powe maps compa isons o each
WM load condi ions pai and equency can be ound in he supple-
men a y in o ma ion ile.
3.3. Analysis and esul s o he dis ibu ion o as equencies ampli ude
ac oss he a cycle angles
sLORETA cu en densi y powe analysis, showed ha WM load
modula ions o oscilla o y ac i i y we e la ges in pos e io a eas o
be a and gamma equency bands and in on al egions o he he a
equency. Thus, we assessed whe he he dis ibu ion o pos e io be a
and gamma ampli udes ac oss he on al he a cycle angles was
modula ed by WM load.
To do his, we used he ini ial 5500 ms epoch s a ing a sample
s imulus onse . Based on he o iginal ol age alues, he LORETA al-
go i hm, as implemen ed in B ain Vision Analyze 2.1 (BVA), was used
o es ima e o each da a poin in he epoch he cu en densi y a he
egions ha showed signi ican WM load- ela ed ac i i y modula ions in
he p e ious analysis (see Table 1). In his s ep, gy i we e used ins ead o
oxels due o he sou ce space in BVA LORETA ans o ma ion being
es ic ed o 2394 oxels a 7 ×7 mm spa ial esolu ion, making i
di icul o ha e a one–one co espondence wi h he oxels o m
sLORETA analyses.
A con inuous complex Mo le wa ele ans o ma ion wi h Gabo
no maliza ion and a Mo le pa ame e c o 5 was applied o hese cu en
densi y es ima es in o de o calcula e hei ime– equency decompo-
si ion be ween 1 and 70 Hz in 30 equency s eps. A e his s ep, o
a oid edge a e ac s, he da a was segmen ed in 3000 ms epochs s a ing
a sample s imulus o se . These epochs we e a e aged o each pa ic-
ipan and expe imen al condi ion.
Fo each expe imen al condi ion and pa icipan a e age, ins an a-
neous he a phase alues we e ex ac ed om he bila e al an e io
cingula e co ex, and igh in e io and middle on al gy i ROIs. In
addi ion, ins an aneous cu en densi y ampli ude alues o be a 1, be a
2, gamma 1, and gamma 2 we e ex ac ed o he ollowing pos e io
ROIs: le usi o m gy us, bila e al cuneus, p ecuneus and pos e io
cingula e, igh middle occipi al gy us, supe io pa ie al lobe, and
pos cen al gy us. No e ha o medial egions showing cu en densi y
powe modula ions by WM load, such as cuneus, p ecuneus, and
cingula e gy us, we decided o analyse bo h hemisphe es due o he
sLORETA es ima ion o signi ican ac i i y being oo close o he
midline. A de ailed lis o he oxels comp ising each ROI is a ailable a
he Supplemen a y In o ma ion ile.
To calcula e he dis ibu ion o as equencies ampli ude along
on al he a cycle, he same p ocedu e as used by Sauseng e al. (2009)
and Pinal, Zu ´
on, Díaz, and Sauseng (2015) was used. Be a 1, be a 2
gamma 1, and gamma 2 ins an aneous cu en densi y ampli ude alues
om each o he a o emen ioned pos e io ROIs we e z- ans o med o
each single epoch, equency band, and ROI independen ly. The pos e-
io ROIs’ z- ans o med cu en densi y ampli ude alues we e hen
g ouped o each on al ROIs’ he a cycle angle and so ed acco ding o
hese ins an aneous he a phase alues. This s ep was done indepen-
den ly o each o he s udied on al ROIs. These he a-phase-so ed z-
ans o med be a and gamma ins an aneous cu en densi y ampli ude
alues we e independen ly a e aged o each on al ROI in o 16 on al
he a phase bins, each bin spanning 22.5◦o a he a cycle o 360◦. The
equi alence be ween bins and he phase angles in deg ees can be seen in
Fig. 3.
As a esul , o each o he h ee on al ROIs, he e a e 16 he a phase
bins wi h hei co esponden cu en densi y be a 1, be a 2, gamma 1,
and gamma 2 z- ans o med ampli ude alues om each o he se en
pos e io ROIs. This p ocedu e was done sepa a ely o each WM load
condi ion using cus om made MATLAB R2016a sc ip s. The numbe o
ials included in hese compu a ions was equa ed be ween g oups and
wo king memo y load condi ions.
Finally, o each pai o on al and pos e io ROIs, he a – phase
so ed z- ans o med cu en densi y ampli ude alues o be a 1, be a 2,
gamma 1, and gamma 2 in each WM load condi ion we e en e ed in o
epea ed – measu es ANOVAs. The ocus o his s udy was exclusi ely
he in e ac ion be ween he wo wi hin – subjec s ac o s: WM load (LL,
ML, and HL) and F on al The a Phase (16 phase bins). The e o e, inde-
penden ly o each pai o on al and pos e io egions, a epea ed –
measu e ANOVA was made wi h F on al The a Phase (16 phase bins)
and WM Load (LL, ML, HL) as wi hin-subjec ac o s o be a 1, be a 2,
gamma1 and gamma 2. The main in e ac ion e ec s o he ou e-
quency bands we e con olled o mul iple compa isons wi h he Holms-
Bon e oni co ec ion; and G eenhouse – Geisse co ec ion we e
applied in all cases he condi ion o sphe ici y was no me . Fo all he
signi ican in e ac ion e ec s, pos hoc compa isons we e compu ed
wi h he Bon e oni co ec ion. The pos hoc compa isons ocused on
con as ing he ampli ude o a gi en as equency a each single on al
he a-phase bin be ween WM load condi ions. Fu he mo e, an associ-
a ed p- alue o 0.01 was used as s a is ical h eshold
Table 1
B odmann a eas, gy i and oxel coo dina es wi h ex eme obse ed -s a is ics om he cu en densi y powe analyses. Fo each SnPM es wi h s a is ically signi ican
esul s, i is lis ed he B odmann A ea (BA) wi h he mos ex eme s a is ic, he label o he gy i showing signi ican oxels in ha BA, he coo dina es in he Mon eal
Neu ologic Ins i u e space o he peak oxels, and he and associa ed p - alues o hose peak oxels.
F equency band Compa ed condi ions B odmann A ea X(MNI) Y(MNI) Z(MNI) Gy us Ex eme - p
The a (4–7 Hz) ML >LL 47 40 40 −5 Righ middle on al gy us 4.23 0.0055
40 30 −5 Righ in e io on al gy us 4.08
HL >LL 25 0 10 −5 Righ an e io cingula e 5.20 0.0001
Be a 1 (13–20 Hz) ML >LL 18 & 30 5 −80 5 Righ cuneus 5.23 0.0008
18 5 −85 0 Righ middle occipi al gy us 5.09
30 20 −65 10 Righ pos e io cingula e 4.95
HL >LL 30 5 −70 10 Righ pos e io cingula e 4.75 0.0007
Be a 2 (20–30 Hz) HL >LL 18 −25 −70 15 Le pos e io cingula e 3.77 0.0156
−20 −70 15 Le cuneus 3.77
ML >HL 20 −35 −40 −25 Le usi o m gy us 3.57 0.0482
Gamma 1 (30–50 Hz) No signi ican esul s we e ound o any compa ison
Gamma 2 (50–70 Hz) ML >LL 7 25 −60 65 Righ supe io pa ie al lobe 3.49 0.0413
HL >LL 7 30 −50 65 Righ supe io pa ie al lobe 3.69 0.0279
25 −55 70 Righ pos cen al gy us 3.50
ML >HL 7 & 31 20 −65 30 Righ p ecuneus 3.45 0.0188
A. Fe n´
andez e al.
Neu obiology o Lea ning and Memo y 183 (2021) 107476
6
Due o he numbe o analyses, only s a is ically signi ican in-
e ac ions be ween he ac o s will be epo ed.
The ANOVAs showed signi ican in e ac ions be ween WM Load and
F on al The a Phase o igh in e io on al gy us he a phase and
gamma 1 ampli ude a bila e al pos e io cingula e co ex (Table 2).
Addi ionally, signi ican in e ac ions we e ound be ween WM Load
and F on al The a Phase o igh middle on al gy us he a phase and
be a 1 ampli ude a bila e al cuneus; and be a 2 ampli ude a igh
pos cen al gy us and igh middle occipi al gy us (Table 2).
Finally, a signi ican in e ac ion be ween WM load and on al he a
phase o bila e al an e io cingula e he a phase and be a 2 ampli ude
alues a igh middle occipi al gy us was ound (Table 2).
Pos -hoc analysis showed he ollowing esul s:
3.3.1. The a phase a Righ In e io F on al gy us
Gamma 1 (30 – 50 Hz)
Gamma 1 ampli ude a bila e al pos e io cingula e was signi ican ly
g ea e a he a p e-peak (bin 14) in he ML condi ion compa ed o he
HL condi ion (Fig. 4A).
3.3.2. The a phase a Righ Middle F on al gy us
Be a 1 (13 – 20 Hz)
Be a 1 ampli ude a bila e al cuneus was signi ican ly g ea e a he a
pos -peak (bin 4) in he LL condi ion han he ML condi ion. (Fig. 4C).
Be a 2 (20 – 30 Hz)
Be a 2 ampli ude a he igh pos cen al gy us was signi ican ly
Fig. 2. B ain a eas showing s a is ically signi i-
can ex eme -s a is ics in he cu en densi y
powe analyses. In each image, i is plo ed he
b ain si es showing he ex eme signi ican -s a-
is ic in he SnPM con as showing s a is ically
signi ican be ween WM load condi ion di e -
ences o he s udied equency bands. Colou
ba s a e based on - alues (see Table 1), wi h
wa m colou s indica ing inc eases o cu en
sou ce densi y powe wi h memo y load and cold
colou s dec eases. No e ha compa isons wi hin
he Gamma 1 equency band (30–50 Hz) a e no
included since no s a is ically signi ican esul s
we e obse ed be ween WM load condi ions.
Fig. 3. Equi alence be ween bins and phase angles in deg ees.
A. Fe n´
andez e al.
Neu obiology o Lea ning and Memo y 183 (2021) 107476
7
g ea e in he HL condi ion compa ed o he LL condi ion a he a ough
(bin 7) (Fig. 4B).
Addi ionally, be a 2 ampli ude a he igh middle occipi al gy us
was signi ican ly highe a he a peak (bin 15) in he LL condi ion han
he ML condi ion (Fig. 4D).
3.3.3. The a phase a An e io Cingula e
Be a 2 (20 – 30 Hz)
Be a 2 ampli ude a he igh middle occipi al gy us was signi ican ly
g ea e a he a pos -peak (bin 3) o he ML condi ion compa ed o he
HL condi ion (Fig. 4E).
4. Discussion
In his s udy, he e ec o WM load on he spec al ac i i y and
loca ion o he a and as (i.e., be a and gamma) equencies was
assessed du ing he main enance pe iod o a DMS ask wi h h ee
di e en le els o WM load o selec he egions o in e es o an
in e egional c oss- equency phase-ampli ude coupling analysis. Thus,
i was, o he i s ime, explo ed whe he WM load modula es he
dis ibu ion o as equencies ampli ude ac oss he a cycle angles.
4.1. Beha iou al esul s
Pa icipan s’ pe o mance was modula ed by WM load ega ding
bo h, pe cen age o hi s and eac ion ime, wi h lowe co ec esponses
and slowe eac ion ime wi h highe WM loads. These esul s we e
expec ed as i has been p e iously obse ed in s udies wi h simila asks
ha inc eases in WM load wo sen pa icipan s’ pe o mance (Al a ez &
Ca anagh, 2004; Vogel e al., 2001).
4.2. Cu en densi y powe esul s
Besides he beha iou al e ec s o an inc easing WM load, he esul s
showed ha du ing he main enance pe iod o he DMS ask, he a and
be a/gamma ac i i y inc eased as a unc ion o WM load, and he la ges
modula ion in WM load was loca ed a di e en sou ces o as and slow
equencies. Besides, we ound ha he be a/gamma ampli ude syn-
ch oniza ion wi h he a cycle phase angles appea s o be modula ed by
WM load.
Fi s ly, ega ding he cu en densi y powe o he a and as e-
quencies, i was ound ha he a ac i i y (4 – 7 Hz) had he la ges in-
c eases wi h WM load in p e on al a eas o he igh hemisphe e
(an e io cingula e co ex, igh in e io on al, and igh middle on al
gy i). These indings a e in line wi h hose om p e ious s udies ha
ound on al he a ac i i y inc eases wi h WM load (Ge ins, 1997;
Jensen & Tesche, 2002; Mel ze , Za e i, Goncha o a, Dis asio, Papa-
deme is, Spence , Spence , & Cons able, 2008). Fu he mo e, g ea e
p e on al he a ac i i y has been ound in associa ion wi h an inc ease
o execu i e con ol demands (G iesmay , G ube , Klimesch, & Sauseng,
2010; Ca anagh, Eisenbe g, Gui a -Masip, Huys, & F ank, 2013).
Taking hese indings oge he , i has been p oposed ha he a ac i i y
has a key ole in WM main enance and execu i e con ol (Mau e e al.,
2015; Sauseng e al., 2010).
Rega ding as equencies, he p esen esul s showed ha as be a
and gamma we e modula ed by WM load in pos e io b ain egions.
Speci ically, be a ac i i y (13 – 30 Hz) had he maximum inc ease wi h
WM load in occipi al a eas, and as gamma (50 – 70 Hz) ac i i y had he
maximum inc ease wi h WM load in igh pa ie al egions.
Be a ac i i y has been ound o inc ease in occipi al and occipi o-
empo al egions wi h WM load du ing he delay pe iod o DMS asks
when memo izing di e en ea u es o geome ical shapes, such as
colo , shape o loca ion (Pal a e al., 2011; Honkanen e al., 2015). This
lends suppo o he idea ha be a ac i i y is ela ed o he main enance
o objec ep esen a ions in WM (Pal a e al., 2011).
Gamma ac i i y has also be ound o inc ease wi h WM load in he
pa ie al co ex du ing S e nbe g (Howa d e al., 2003) and DMS asks
(Pal a e al., 2011). In his ega d, gamma in ol emen in WM has been
equen ly associa ed wi h he main enance o isual ep esen a ions in
WM (Jokisch & Jensen, 2007) and WM capaci y (Tallon-Baud y e al.,
1998). Fu he mo e, indings o as gamma ac i i y in he pa ie al
co ex du ing WM main enance link his ac i i y wi h inc easing
a en ional demands (F ies, 2009; Pal a e al., 2011).
These esul s, howe e , con adic he s udies ha ha e obse ed
dec eases in be a and gamma ac i i y on spa ial WM asks when he
ocus is on memo izing he s imulus loca ion (Poch e al., 2014; P os-
ko ec e al., 2018; 2019). In e es ingly, when analyzing he ac i i y
ela ed o he memo iza ion o he di e en s imulus ea u es, Honkanen
e al. (2015) ound ha as ac i i y inc eased wi h WM load when
pa icipan s we e memo izing shape o colou bu ailed o ind a WM
load e ec when pa icipan s memo ized loca ion alone. Inc eases in as
ac i i y when memo izing dis inc symbols such as le e s (Howa d
e al., 2003; Michels e al., 2010) o i egula shapes (Honkanen e al.,
2015; Mo gan e al., 2011; Tallon-Baud y e al., 1998) seem o be
common. E en hough he pa icipan s in he cu en s udy had o de ec
whe he a do in he domino ile had changed i s loca ion, i is possible
ha , a leas o highe WM loads, hey elied on he o e all shape
pa e n o med by he do s in he h ee iles, a he han he indi idual
loca ion o each do . This ac would explain why pos e io inc eases in
be a and gamma ac i i y wi h highe WM loads we e ound, since hey
would be p ocessing i egula shapes a he han he do s’ loca ion pe
se.
Addi ionally, i was ound ha in wo speci ic b ain egions, his
modula ion o as equency oscilla o y ac i i y by WM load was
e e sed: as be a and as gamma dec eased wi h highe WM load, wi h
he la ges dec eases in he le usi o m gy us and in he igh p ecuneus
espec i ely.
Al hough inc easing gamma ac i i y has been p e iously obse ed in
he le p ecuneus du ing he manipula ion o ea u es o wo s imuli in
WM (Mo gan e al., 2011), he p esen ask only equi ed pa icipan s o
main ain wi hou men ally manipula ing any ea u e o he s imuli (i.e.,
s imulus’ con igu a ion). I could be he case ha gamma ac i i y in he
p ecuneus is speci ically associa ed wi h mo e complex WM manipula-
ions han hose demanded in he p esen s udy. Thus, in he HL con-
di ion, i could be p e e ed o alloca e p ocessing esou ces o he
supe io pa ie al lobe and igh pos cen al gy us, whe e inc eases o
gamma ac i i y wi h WM load we e ound, while ac i i y in he igh
p ecuneus migh be no longe equi ed, allowing a educ ion o gamma
Table 2
Signi ican in e ac ions be ween WM Load Fac o (HL, ML, and LL condi ions) and F on al The a Phase Fac o (16 phase bins) on be a and gamma no malized cu en
densi y ampli ude alues. F (wi h deg ees o eedom in pa en hesis) and p- alues a e indica ed alongside he on al and pos e io egions o he a and be a/gamma
equencies, espec i ely.
F on al egion F on al egion equency band Pos e io egion Pos e io egion equency band F (30/720) p
Righ in e io on al gy us The a (4–7 Hz) Pos e io cingula e (bila e al) Gamma 1 (30–50 Hz) 1.859 0.004
Righ middle on al gy us The a (4–7 Hz) Cuneus (bila e al) Be a 1 (13–20 Hz) 1.890 0.003
Righ pos cen al gy us Be a 2 (20–30 Hz) 1.797 0.006
Righ middle occipi al gy us Be a 2 (20–30 Hz) 1.816 0.005
An e io cingula e (bila e al) The a (4–7 Hz) Righ middle occipi al gy us Be a 2 (20–30 Hz) 2.235 <0.001
A. Fe n´
andez e al.
Neu obiology o Lea ning and Memo y 183 (2021) 107476
8
ac i i y o be obse ed.
Summa izing he p esen s udy indings, inc easing he a ac i i y
wi h WM load in p e on al a eas suppo s he in ol emen o he a
ac i i y in WM main enance and execu i e con ol. Rega ding as e-
quencies, highe be a ac i i y wi h inc easing WM load in a eas ela ed
o pe cep ual isual p ocessing, combined wi h gamma ac i i y in-
c eases in a en ion– ela ed egions may indica e ha , o his ask, be a
ac i i y was enough o isual ep esen a ions, while pa ie al as
gamma ac i i y e lec s he a en ional demands o he ask. This
inc ease o as ac i i y in highe WM load condi ions is in line wi h ha
desc ibed in p e ious s udies when p ocessing and memo izing global
shapes, e en hough some spa ial in o ma ion was equi ed o sol e he
p esen ask. These esul s highligh he need o ake in o accoun he
me hodological aspec s o he ask, as well as he memo izing s a egies
used by he pa icipan s in u u e s udies o WM load e ec s on b ain
ac i i y.
Fig. 4. Pai wise compa isons o as equencies no malized ampli ude in he h ee memo y load condi ions ac oss he he a cycle angles a he igh in e io on al
gy us ( IFG), igh middle on al gy us ( MFG), and an e io cingula e co ex (ACC). Fo each as equency and egion, i is indica ed in he able on he igh he
ask condi ions showing signi ican di e ences, he bin o he a equency band whe e i is localized, and he associa ed p- alues. In he line g aph, he as e-
quencies’ mean no malized ampli ude alues a e shown (S anda d E o -SE- wi h ba s) o each egion. The on al he a cycle bins showing signi ican di e ences
a e g ey shaded. A) he a phase a IFG and gamma 1 no malized ampli udes a bila e al pos e io cingula e. B) he a a MFG and be a 2 ampli udes a igh
pos cen al gy us; C) he a a MFG and be a 1 ampli udes a bila e al cuneus; D) he a a MFG and be a 2 ampli udes a igh middle occipi al gy us; E) he a a ACC
and be a 2 ampli udes a igh middle occipi al gy us.
A. Fe n´
andez e al.
Neu obiology o Lea ning and Memo y 183 (2021) 107476
9
4.3. Dis ibu ion o as equencies ampli ude ac oss he a cycle angles
esul s
Las ly, ega ding he dis ibu ion o pos e io as equencies
ampli ude ac oss on al he a phase angles, i was ound ha he dis-
ibu ion o pos e io be a and gamma ampli udes in he cou se o he
phase cycle o on al he a is in luenced by WM load, p o iding no el
e idence abou WM load e ec s on in e egional c oss- equency phase:
ampli ude mechanisms.
Speci ically, g ea e ampli udes o be a ac i i y a he bila e al
cuneus and igh middle occipi al gy us as well as highe gamma ac i i y
a he bila e al pos e io cingula e we e obse ed o lowe memo y load
condi ions when compa ed wi h highe memo y load condi ions du ing
o nea on al he a peak. Con e sely, g ea e as be a ampli ude a he
igh pos cen al gy us was obse ed du ing he a ough a MFG in he
high memo y load condi ion compa ed o he low memo y load condi-
ion. These esul s e idence ha WM load modula es communica ion
be ween co ical a eas.
Following he model by F ies (F ies, 2005), i is expec ed ha
op imal neu al communica ion is e lec ed in a concen a ion o as
equencies ac i i y a ound speci ic angles o he a phase. The e o e, o
demanding si ua ions, he p e e ed phase angle will be he a ough
and, he e o e, ideally: i) g ea e ampli udes in he highe load condi-
ions (compa ed o lowe loads) should be obse ed du ing he a ough;
ii) as well as g ea e ampli udes in he lowe load condi ions (compa ed
o highe loads) du ing he a peak.
Suppo ing he i s p edic ion based on he model, as be a ampli-
ude a he pos cen al gy us was g ea e du ing he a ough a he
MFG o high han low WM load. Acco ding o F ies (2005) model, his
coupling occu s a he conside ed op imal window o communica ion.
The pos cen al gy us is a soma osenso y egion. Howe e , i has also
been ound o play a ole in a en ion (Shulman, As a ie , McA oy,
d’A ossa, & Co be a, 2007), as well as o be pa o a pos e io a en-
ional ne wo k (Tomasi, E ns , Capa elli, & Chang, 2006). Fu he , i s
lesion has been shown o impai a en ional skills (Bajaj, Dailey, Rosso,
Rauch, & Killgo e, 2018). Gi en ha he MFG is hough o be ela ed o
he main enance o in o ma ion in WM, especially in non- e bal asks
(Daniel, Ka z, & Robinson, 2016; Linden, 2007), i could be a gued ha
o highe memo y load, MFG and pos cen al gy us es ablished
communica ion du ing he model’s op imal window o communica ion
as a e lec ion o he inc eased demands on a en ional con ol o e WM
con en s.
Rega ding he second p edic ion, in he cu en s udy, he pos e io
cingula e has been ound o communica e du ing a non-op imal window
o communica ion (i.e., on al he a peak) wi h he IFG, an a ea
known o be implica ed in execu i e con ol and inhibi ion in WM
(Kasaha a e al., 2011). The pos e io cingula e co ex is conside ed pa
o he de aul mode ne wo k (And ews-Hanna, Reidle , Sepulc e, Poulin,
& Buckne , 2010) and has also been ela ed o spa ial p ocessing and
spa ial na iga ion (Rolls, 2018; 2019). Speci ically, i has been obse ed
o play a ole in he isual p ocessing do sal s eam, ela ed o spa ial
in o ma ion (K a i z, Saleem, Bake , & Mishkin, 2011). Since he ask
used in he p esen s udy equi es he main enance o he s imuli’ global
con igu a ion, i may be he case ha he e was a minimal demand o
communica ion o alloca e cogni i e con ol esou ces on spa ial
p ocessing- ela ed a eas du ing low load ials.
Likewise, he igh middle occipi al gy us is a isual p ocessing a ea.
S ill, i has also been ound o be ela ed o isual memo iza ion s a-
egies. Fo ins ance, i has been obse ed ha simula ing a lesion on his
a ea wi h ansc anial magne ic s imula ion impai s digi span ask
pe o mance in pa icipan s who use isualiza ion s a egies (Hilbe
e al., 2019). This a ea appea s o communica e du ing non-op imal
windows wi h bo h he MFG and he ACC, which a e ela ed o non-
e bal main enance o in o ma ion and o cogni i e con ol, espec-
i ely (Daniel e al., 2016; MacDonald, Cohen, S enge , & Ca e , 2000).
This non–op imal communica ion has been ound o lowe WM loads in
bo h cases, ollowing he same pa e n as indica ed abo e, whe e lowe
WM loads demand less cogni i e esou ces, and hus he e is a lesse
need o on al con ol o occipi al a eas.
I is no iceable ha mos a eas ha yielded a signi ican esul a e
la e alized o he igh hemisphe e. I is belie ed ha coo dina e spa ial
ela ion p ocessing is la e alized o he igh hemisphe e and ha his
pa e n is also gene alized o isual wo king memo y ( an de Ham, an
Wezel, Oleksiak, & Pos ma, 2007; an de Ham, Pos ma, and Laeng,
2014).
In conclusion, he p esen s udy esul s o e ed no el e idence
showing ha , du ing he main enance pe iod o a DMS ask, WM load
modula es in e egional communica ion be ween on al and pos e io
a eas. This shows ha in e egional communica ion depends on he
cogni i e demands equi ed by he ask o a success ul pe o mance,
wi h WM load ac ing as a swi ch ha enables o hinde s he op imal
communica ion be ween on al and pos e io a eas depending on he
need o on al con ol in isual and a en ion p ocessing egions.
CRediT au ho ship con ibu ion s a emen
Alba Fe n´
andez: pe o med he da a analysis and in e p e a ion
unde he supe ision o Diego Pinal and w o e he i s d a o he
manusc ip . Diego Pinal: designed he s udy and he DMS ask along-
side Fe nando Díaz and Mon se a Zu ´
on, supe ised he da a analysis
and in e p e a ion, and p o ided c i ical e isions o he manusc ip .
Fe nando Díaz: designed he s udy and he DMS ask alongside Diego
Pinal and Mon se a Zu ´
on, and p o ided c i ical e isions o he
manusc ip . Mon se a Zu ´
on: designed he s udy and he DMS ask
alongside Fe nando Díaz and Diego Pinal, and p o ided c i ical e isions
o he manusc ip .
Acknowledgemen s
This s udy was suppo ed by g an s om he Spanish Go e nmen ,
Minis e io de Economía y Compe i i idad (PSI2014-55316-C3-3-R;
PSI2017-89389-C2-2-R), wi h FEDER Funds; he Galician Go e nmen ,
Conselle ía de Cul u a, Educaci´
on e O denaci´
on Uni e si a ia, Axudas
pa a a Consolidaci´
on e Es u u aci´
on de Unidades de In es igaci´
on
Compe i i as do Sis ema Uni e si a io de Galicia: GRC (GI-1807-USC);
Re : ED431-2017/27, wi h FEDER unds. D. P. was also suppo ed by he
Po uguese Founda ion o Science and Technology (FCT) h ough he
ellowship SFRH/BPD/120111/2016.
Decla a ion o Compe ing In e es
Au ho s decla e no con lic o in e es .
Appendix A. Supplemen a y ma e ial
Supplemen a y da a o his a icle can be ound online a h ps://doi.
o g/10.1016/j.nlm.2021.107476.
Re e ences
Al a ez, G. A., & Ca anagh, P. (2004). The capaci y o isual sho - e m memo y is se
bo h by isual in o ma ion load and by numbe o objec s. Psychological Science, 15
(2), 106–111. h ps://doi.o g/10.1111/j.0963-7214.2004.01502006.x.
And ews-Hanna, J. R., Reidle , J. S., Sepulc e, J., Poulin, R., & Buckne , R. L. (2010).
Func ional-ana omic ac iona ion o he b ain’s de aul ne wo k. Neu on, 65(4),
550–562. h ps://doi.o g/10.1016/j.neu on.2010.02.005.
Axmache , N., Hensele , M. M., Jensen, O., Wein eich, I., Elge , C. E., & Fell, J. (2010).
C oss- equency coupling suppo s mul i-i em wo king memo y in he human
hippocampus. P oceedings o he Na ional Academy o Sciences, 107(7), 3228–3233.
h ps://doi.o g/10.1073/pnas.0911531107.
Baddeley, A. (1998). Recen de elopmen s in wo king memo y. Cu en Opinion in
Neu obiology, 8(2), 234–238.
Baddeley, Alan (2003). Wo king memo y: Looking back and looking o wa d. Na u e
Re iews Neu oscience, 4(10), 829–839. h ps://doi.o g/10.1038/n n1201.
A. Fe n´
andez e al.
