Attentional propagation of conceptual information in the human brain

The v is u al env iro nment con ta in s m ore in forma tion th an w e are a ble t o ev alua te. To ac t adaptiv ely, we p r ioriti s e s ome inf ormati on s ourc e s o v er othe r s u s ing a se t o f mec hani sm s colle ctivel y re fer r e d to a s selective attention ( De simon e & Du nca n, 1995; Egeth & Ya n t i s , 1 997) . Ea r ly work charac te r i sed att entio n in no w-famou s met aphor s , s u g ges ting a s p otlig ht o r fil ter t ha t act s in re tinot opic s p ace t o enhanc e sen s ation a nd perc eption (Pos n er, S nyder , & Davids on, 19 80; Br oadbe nt , 1 958). Guid ed by the s e id ea s, neu ro scie n t i f i c inv e stig ation has foc u s s ed c lo s e ly on iden ti fica tio n of the sys t e m s a nd algorithm s th at impa ct se nsory cort ex (Maun sell , 2015; D e simon e & D unc an, 19 9 5; Rey nold s & H e eger , 200 9) and on cha rac t eriz a tion o f the brain n e t w ork s t h a t con trol t he se m ec hani sms (U ng erl eide r, 200 0; C o rbe tta , P a tel , & S h uman, 2 008; Hüer, S a xena , & Tr e ue , 202 4). The e me rgent model sugg e st s tha t stra tegi c selec t i on b egi ns w ith the de riv ation o f schem a s a nd templa te s fr om conc eptual a nd s emantic knowl edg e. Th es e are s tored in wo r k ing mem or y at f ro ntal brain si te s and tr a n s la ted t h r oug h t he f ron tal ey e field s and po s te rior pa rie ta l cor tex into bia s es of s e n s o ry activi t y (De s imo ne & Dunc an , 1 995; Mille r & C o h e n, 2001). .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 4 Res earch o n the n europhy s i ology o f at t e nt i on ha s thu s focu s s ed clo se ly on how conc ept s and ca tegori e s are u s ed to g uide se ns atio n a nd percep tion. In contr a st, ou r unde rs t a nding of th e r e ver se infer enc e - of h ow att entio n sup por ts th e activ ation and con struc tion o f c oncep t s - i s r elative ly poo r. Tha t is, we k now that a t t entio n plays a k ey r ol e in in forma t i on gath ering and th e g enera tion a nd ac tivation o f high-l ev el k nowled ge (Go tt l i eb, Oud eye r , Lope s , & Ba rane s , 201 3; Bu zs ák i, 201 9 ). And , whi le ther e is d eb a te abo ut th e d egr ee t o w hich conc epts a r e grou nded i n s e n s o r imotor system s (Bar sa lou, 20 16) , we know that t he sema ntic r e pre sen tati on s tha t eme r g e in fr ont al a nd tempo ral co r tex ab strac t fr o m pe rceptu a l exp er i en ce ( F re edman, Ri e senhub er, Pog gio, & M iller, 2 001; Rig o tt i et al . , 2013, Fairha ll & C aramaz z a , 2013; A c unzo, Low , & Fai rha ll, 202 2) . We p re s u m e that a tt enti on se rves the pu rpo se o f ac tiva ting and c on s t ructing th e s e conc ep tual repr es ent atio ns t o sup por t po s t - perce p tual c ognitio n, but we k now l ittle of how t hi s un fol d s i n th e brain . To a ddre ss thi s, in the curre nt p aper w e i denti fy and c harac teri s e neu ral rep r e s en ta t io ns o f abs tr a c ted i n f o r ma t io n th at vary a s a fu n ction of the qual i ty of a tt entio na l s e lec ti on. This a pproac h has b ee n ado pte d i n earli er w ork. Fo r e xampl e, re s u l ts s how tha t the s em a ntic c a tegory of a vi s ua l st i mulu s i s enco de d in pa t te rn s of fMR I a c t i vity in ventral a nd do r s a l visu al cort ex (Hax by et a l. , 200 1; Jeong & Xu , 2016 ) a nd tha t th e se r epre se nt a t i on s v ary a s a func tion o f the i r beha vi our a l relev anc e , co ns i st ent with a n e ff ec t o f at t e n t i on (Se idl , Pe ele n & Kastne r, 2 012; Hic key & P e e len, 2 015; Barbar o, P e el en, & Hic key, 2018) . How ev er, thi s pa tt ern ha s la rge ly bee n id enti fie d in brain are a s that a re l ikely to enc ode vi s ua l f ea ture s , makin g it unc lear th at it r efl ec t s a n e f fec t on categ o r y informa tion rath er tha n on perc e ptio n of the v is ua l fe atu re s tha t happ en t o b e sh ared ac r oss ca tegory ex ample s ( Brac ci & Op de B e ec k, 2016; Wat son, H artl ey, & An drew s , 2 0 14; Ric e, Wa t son, Hartl ey, & And rew s, 20 14). To c ircumv en t thi s i s s ue, a s e p ara te r e s e a rch li ne ha s emplo yed mul t i - moda l stimul i tha t ac tivat e th e same con c ept - for exa mple, b y pre sen ting a n ima ge of an a ppl e a nd the w or d ‘a pple ’ . This h as id en t i fi e d ca t e gory informa tion in ar ea s tha t do no t car ry perc eptu al signal s , such a s an teri or t emporal a n d pr efron t a l co rtex (Rigo tti et al ., 2013; Bind er et al . , 2009; Fa irhall & Ca rama z za , 201 3; Acunzo , L ow, & Fa irh all, 2 022) . Howev er, the rol e o f a tt e n t i on in t he ac tivation o r c ons t ruc t i on o f the s e rep re s e n t a t i on s rema i ns uncl e ar . As a r e sult , the re ar e hol e s i n ou r unde rs t a nding o f t he n eural me ch ani s ms that s upport in for m a tion ga t he r i ng . We pr e sume th at in forma tion s el ecte d thro ugh t he d eploy ment o f a tte ntion a ctiva t e s or co nstruct s ne ural r epre s ent ation s o f co n cept s and c at egorie s , but w e hav e no t ve r ifi ed thi s, and w e do no t k now w here in the brai n thi s tak e s pla ce . A r e a ll s uch high-lev el repr e sen t ation s s en s i tiv e to att entio n, o r i s th e e f fe ct o f at tent ion l im ited to di scr et e brai n n etwork s? A nd what doe s t his me a n for ou r under s t a nding o f t h e func tiona li t y of the se area s? H e re, we u se no vel me tho dology and .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint Fi g ure 2 – Res ul ts fro m E EG. a.) E RP res ult s fr om Ex perim e n t 1. Th e N 2 pc is evi den t a s a neg a ti vit y in t he p ost erior wavef o r m record ed o v er cort ex c ontr a l at e ral to a n at te n d e d s ti mul us. Th e bl ue b ox ide n tifi e s a 1 2 5 ms in te r val c e n tere d o n t he cro ss -co ndit io nal pea k of t he N2pc . T he t o pogra p h ical p lo t ref l ect s th e vol ta g e diff ere nc e b e t w een co nditi o ns w hen th e tar g e t w a s i n t he l e ft ve rs u s rig ht v is ua l fi eld; t he N2 pc ex press es a s a ne g a ti vi ty i n t he le f t he m is p here bu t a p o s it iv i ty i n t he ri ght hemis p h ere du e t o t he s ubt ra c tion. Con tralatera l an d i psil a tera l E RP s r e fl ect me a n s ig na l acros s t he 2 s ets o f 3 lat e ral and po steri or electro des ide n tifi e d in th e to p og r ap h y b y la r ger m ar k er. b.) I ps ilateral- m i n us-co ntralat eral dif fe r ence w a ves fro m E xperi m e nt 1 . T h e N2 p c is e vi d ent as n e g a t iv e def l ect io n b e gin ni ng a ro un d 1 7 5 ms pos t-s ti mul us. Sha din g here a n d in o t her p a n e ls in di c a t es b o ots tra ppe d SE M. c.) T i m e-res olve d c l a ss i f ic a ti o n of th e tar ge t a nd dis tract or o bject c ate go r y in Ex peri m e nt 1 . A s t here were 4 c l ass e s , c h a nc e perfor ma nce w as 2 5% . d.) ERP re s u l ts fr o m E x pe ri me nt 2. T hes e reflect s ig nal at t he lat e ral pos t e ri or electr od es i den tified in w hit e i n t h e t opo gra phy pres en ted i n pa n el A . e. ) Ipsi la t eral- m i n us-co ntra la t eral di ffe r ence w a ve s fro m Ex peri men t 2 . f.) T i me -r e s o l v e d clas sificati on of t a r ge t a nd di s t r a c to r o bj e c t ca te go ry in E xpe r i m ent 2 . .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 6 ana lys e s t o iden ti fy the e f fect o f a tten tio n o n informa tion a b out th e s e man tic cat ego r y of visu al st i muli . Our appro ac h relie s on i d enti fica tio n of tr i al -by -tri al c ovaria nce b e t w e en a wel l- ch aracte ri sed EEG i nd ex of spati a l at ten ti on – t he N 2 pc (L uck & Hilly ard, 1994 ) – a nd an inde pen den t mea s u re o f seman tic ca tego ry informa ti on derived u s in g mac hine l e arning o f ne uroimagi ng data . Ex periment 1 e mploy s hig h-d en s i t y EEG i n the iden tific ation o f thi s in form ation to a llow f o r p rec is e tempor al r e solu tion . Experime nt 2 empl oy s conc ur r ently -r ec o rded EE G a nd MRI to provi de pr eci se s p at i a l lo c a li z a t i on.

Par tic ipant s in bo th ex perim ent s ( n = 44, n = 31 ) were a ske d to compl e te th e sam e ta sk w ithout movi ng their ey e s. Thi s inv olved r ep or tin g , via butt o n pr ess , the c a tegory o f a ta rg et obje ct (Fig 1a) that ap pe a red a t a cued loca tio n (Fig 1b ), whil e ign or in g a se t of n on- targ et s th at i nc luded v isually deg r a ded o bjec t s a nd a sing le r ecog niz a b le di strac tor . Objec ts we re sel ec ted from a s e t of image s that w ere visu ally hete rogen e ou s within ea ch c atego ry. Pa rticipa nts c omple ted th e t a s k quick ly (E 1: 1076 ms , 185 SD ; E2: 108 8 ms, 240 SD ) a nd ac curate ly (E1: 87 .6%, 1 0.6 S D ; E 2: 89.9% , 7.0 S D). In bo t h expe rime nts , w e mea sur ed the q u ality of a tten t i o nal proc e ssing u s ing th e N2pc c omponen t of th e v is u al ev ent -rel at ed pot enti al (ER P ; Lu ck & Hilly ard, 1994 ). The N2pc emerg es ov e r po s t erior co r tex contr ala te r a l to th e loc ati on of a n att ended ob jec t (L uck, G i rel li, Mc Dermo t t, & Fo rd, 199 7; Hick ey, Di Lollo, & Mc D o nald , 20 09). I t in c r e a s e s i n amplit ude a s circu m s tanc e s r e quire a g r e a te r inv es tm ent o f at t e n t i on al re sou rce s (B ri sson, R obi ta ille, & Jolic o eur, 200 7; R a shal et a l. & Bo ehler , 2022 ) and is the re fore u s ed a s an index o f the d epth o f a t t entio n al proce ssing (L u ck , Gir e lli, McDermo tt , & For d, 1997 ; Luc k & Ford, 1 998). We tr ack ed th e em er g e nce of ca t e gory i nforma tion in th e brain u s in g lin ea r di s c r iminant a naly si s (LDA ) to c la ssi fy neuroimag ing da ta . We tr a in ed mode l s t o cla s si fy the obj ec t c at e gory of the t a r g e t or dis tr a c tor ba s ed on no r ma l ise d pa tte r ns o f evoke d bra in a ctivity , s ub sequ e nt l y te st i ng th e se model s o n r e t a ined d a t a to cal cula te mo del c las s i fi ca tion a cc ur a cy. Measuring attentional selec t ion and ca t e gory inform ation in E EG In e ach experim en t a l tr i al , the targe t obj ect c ould a ppe ar t o the l e ft or rig h t o f a c e ntral fixa t io n mark or directly ab ove or below it ; when t he ta rget wa s pr es ent ed la te r a lly, t he s i n gle rec ognizabl e dist r a ct or wa s pre se nte d on the ve rtica l, and v ice versa (Fig 1b). Thi s de s ig n wa s a do pt e d a s stimul i on the ver tical meridi an of the vis ual fi el d ini t i ally crea te a ctivi ty of equ al magni tude in both c or t i cal hemi s ph er es , and th er e fo r e a tt e nt i on al s elec tion of t he se stimuli d oe s no t g e nera te th e l at eraliz ed .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint Fi g ure 3 – a .) Anal y s i s of E E G da ta i n Ex peri men t 1 . Trials were re p eat edl y pa r t i t io ned i n t o trai nin g a nd t est s e ts . Cl a s s ificati on a c curacy a nd mean N 2p c am pl i t ud e w ere m e as ured f or each tes t s et, re s ulti n g i n t wo de pen den t me a sure s ( ide ntif i e d in b o l d). b.) Ana l ys is o f M RI a nd E EG da ta i n Ex p eri ment 2. T he i m a gin g mo dalit ie s w e re lin ked d u r in g tr i al partit io n i n g , suc h th at t he sa m e trials a p p ear i n bo th MRI a nd E E G trai nin g an d tes t set s . T hi s res ul te d i n t h r e e de pen de n t m ea s u res (i den tifi e d i n b ol d ). c.) To li nk EE G and MRI res ult s in Ex peri m e nt 2 w e correlat ed M R I cla ssifica ti o n acc u racy wit h N 2 pc am p l it u d e o b s e r ved ac ross t h e set o f 1 1 4 0 s a m plin g it eratio ns. This occ u r re d f or e ac h s e a rc hl i g ht s ph ere. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 8 EEG v oltage unde rlying N 2 pc (Woo dman & L uck , 2003). By pre se nting stimuli on t he ve r tic al w e we r e able to i s ola t e th e N2pc elici t e d exc lusively by t he ta r g e t or di st r ac to r ( Hic ke y , McDonald , & The euwe s, 200 6; Hick ey, Di L ollo , & Mc Dona ld, 200 9) . The N 2p c eme r g ed p rominen tly in both e xperiment s (Fig s. 2 a, 2d ) a nd wa s la r g er in amp litude for targe t s than r eco gnizabl e di s tr actor s (Fig s 2b, 2e ). I t wa s p rec ed ed b y earli e r , po s i tiv e-p olari ty l ate ral vol t a ge tha t re flec t s a n imbal ance in s en s o r y activ i ty betw ee n the v i s u al hemi s ph ere s. In o ur st i mulu s di spl ays, ther e i s a dif fe r e nce i n the c omplexi t y and spa tia l f r e qu ency of t he c omplet e objec t, pr es ent ed on on e s i d e of the scr e en, and th e un rec ognis able deg rade d o bjec t , pr esen ted on the ot her (Fi g 1b). T hi s kind of imbal an ce c r e ate s s ho rt -la tenc y lat eral i t y in the vis ual E RP (i e. P P N; Pome r l e au e t al . , 2014). The in s en sitivity of thi s e f f e ct to th e be haviou ral r e l eva nc e of t h e evok ing st i mulu s di sting ui she s it fr om t h e s ub seq ue nt N 2pc . Targe t and di s tr a ct or ca te gory info rmati on was iden ti fie d in th e E EG u s i ng cl assi fi c ation an aly si s. Fo r eac h s ample in th e epo ch ed d at a, an LDA model was t r a ine d on r e sul t s ob ser v ed i n a ~85 ms wi ndow c entred on that ob serva tio n. A c ros s- fold v alida t i on scheme – d e scribed fu r the r bel ow – wa s empl oyed to t e st the model in ea ch of th es e w indowe d s a mple s, gen e r a t i ng a t im e -cou rse o f cl assi fic atio n ac curac y. A s ill u s t rat ed in F i gure 2c , w hen the targe t a ppe are d la te r a lly, t he cat eg or y co uld be rel ia bly cla ssi fied from ap p r ox imately 215 m s po st - s timul u s in Expe r im e nt 1 (Fig 2c) and from 240 ms i n Expe r i men t 2 (Fi g 2f ). Di strac to r c la s s i fica tion al so emerged in Ex p erimen t 1, thoug h later a nd with red uc ed ma gnitud e a nd e x t e nt . Import antly, we d e s ig ne d the e xperimen ts such th at c la s s i fic atio n of objec t c at egory could not lev erage the EEG va r ia nc e under pinning N 2 pc. In all c la s s i fi catio n a naly se s, mode l creation and va lidation wer e ba sed o n se ts o f trial s th at con t a i n the sam e nu mber s o f rel e vant v ehicl es , t o ol s , food s, a nd animal s, a nd r oughly the sam e numbe r o f le ft - and righ t-l a t e ral ized sti muli within each of the se ca t e gori e s. The loca tion o f th e objec t – and t h u s the l ate rality o f t h e N 2 pc i n left o r r i ght vi sual co r tex – provi de d no i nfo r ma t i on abou t i ts ca t e gory . Coupling varianc e in N2p c to varianc e in EE G -de r iv ed ca t eg or y infor mation We le ver aged nat ural va r i a nce in th e qua lity of a tten tion al s elec t i on to ide nti fy cova r i ance in N2pc and categ o r y cla ssi fica tio n ac curac y. Thi s v arianc e stem s in par t from chang e s in p artici pan t vi gilanc e, ta sk e ngage men t, and f a t i gue, bu t a l s o r e flec t s confu si on a nd th e mis all oc ation o f att entio n to di s t racto r obje ct s. O ur ex pe ctation wa s th at wh en the qua li ty of a tt e ntiona l d eploy ment to the targe t w a s high, t hi s w ould c au se emerge nce o f in for m a tion ab out th e at t e nded obje ct in th e neural s ig nal . .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint F ig u r e 4 – Lin ki n g N2 p c and EE G ca t e g o ry i n fo r m at io n in E xp e r im ent 1 . a . ) Wi thi n -pa r t i c i pa nt r e su l ts f r om 3 exam p l e p art icip ants . A c ro ss th e 1 1 40 s ampl i ng it e ra ti ons , th e ta r get- el icit ed N 2pc o bs erved in a gi v e n s am pl e pre di c ts tar g e t clas sifica tio n acc u racy in t ha t s a m ple. b .) Lin ea r fi t of tar ge t - e licit e d N 2 pc a nd t arget c l ass i f i c ati on acc ura c y f or e ac h pa rt i c ipan t, w i t h me a n re la ti o n sh i p i de nt i fi ed i n red. c.) Lin e ar fi t of di stract or- elicited N 2pc and t a rg et cl a s sif icatio n a c curacy f or ea c h part icip a n t, wit h m ean re l at io n s hip i d e ntif i e d i n re d . To te st t h i s, w e iter ativ ely re s ample d th e data i n a boo t strap r o u t in e, bu ildi ng c las s i fic atio n mode l s and me a s uring th e N2pc i n eac h in s t a nce ( F ig. 3 a) . For e a ch pa r tic ip ant, targe t - la te r al and di stract or - later al tri al s we r e s e p ara ted in to 20 chunk s , e ach con taini ng an e qual nu mbe r of t r ia l s where t he cla ssi fied ob jec t wa s a food , v ehicl e, ani ma l, or tool . T he c hunk s were a ll oca t e d r epea tedly into two s ub se t s. The l a rger s e t – con sti tuting 17 chunk s or 8 5% of trial s – w a s u s e d to trai n a c las s i fic atio n model tha t i den tifi ed th e ta r g e t cat eg or y . The smal l er s e t – con sti tut i ng t he rem ai ning 3 chunk s or 15% of t ria l s – wa s u sed bo th t o te s t the model and to me a s ure the ta r g e t- o r dist ractor -elic ited N2 p c . For eac h o f the targe t- la te r a l and di s tr actor - lat eral con dition s , we iter at ed thi s pr oces s o f s a mpling and mode l build ing 114 0 time s, r efl ectin g all possible combin a tion s o f 3 chu nks ta ken fr om 20 . The cla ssi fica t i o n ac curacy illus tra ted in F igur es 2c and 2 f re flec ts the mea n of the se v alue s ac r o s s itera tion s and sub se quen tly pa r tic ip ant s . In ea ch ite rati on we me a s ur ed t h e N 2 pc as mea n volt ag e in a 125 ms inte r v al c entr ed at the p e ak (a s ide nt i fi ed from da ta c ollap se d a cro ss ite ration s , c ondition s , and pa r tic ip ant s; 173 – 298 m s) . We me a s ur ed me an cl a ssi fic atio n ac curac y in t hi s same in ter v al, r e s ulting in pa i r e d s e ts of 11 40 N2pc a nd cl assific ation ac cu rac y va lues for eac h a naly s i s a nd .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 10 partic ipan t. As illu s tr a t e d in Fi gur e 4a, wi thin- par tic ipant r e s u lts c ommonly show e d a r e lati on ship betwe e n targ et -elic i ted N2pc and targe t c las s i fi catio n a cc ur a cy, with ne ga tive-v ol tage N 2pc ampl itude p redic ting an impr ovemen t in targe t cl a s si fic a tion. Th e mean r e l atio ns h ip is plot ted in Fi gure 4b a long side r e sult s from e ach o f t he 44 pa rtic ipant s . One int erpr et ation o f th is p at tern i s th at robu s t a t t entio n al s elec t i on o f the targe t – r e fl ect ed i n large r ta r g e t - e licite d N 2p c – c au sed the E EG t o carry more in fo r m ation about the t a r g et ca teg o r y . Howev er, an a l tern ative i s th at bo th N2p c and ta r g e t info rmatio n c ovaried a s a fu nction of partic ipan t ar ou sal or t a sk en gage men t without an y more di rec t r e la tion ship . W e c onducted a secon d analy s i s to pro be thi s al t e r na t i ve . Here, w e r elat ed t a r g et in form ation to d ist r a cto r- elici te d rath er tha n targ et -el ici ted N2pc . If the N 2pc and c ategory in for m a tion bro adly cov ary, this a na lysi s should id e nti fy the sam e rel a tion s h ip a s desc ribed ab ove. H ow eve r, a s i llu st r ate d in Fig ur e 4c, t h e mag nitude o f di str actor - elic ited N2pc i n f act predi c ted a d ecrea s e i n targ et cl a ssifi c ation ac curac y. We u sed hie r a r c hica l lin e ar mod elli ng of trial -w ise EEG and m odel as s e ssment to sta t i s t ic ally ev aluat e the s e pa tter ns (Hicke y, Grign oli o , Muna singhe , & Ac unzo, unde r revie w) . An ini t i al mod e l of cl assi fic atio n ac curac y in cluded a single ra ndom ef fec t fo r the p er -par t i cipa n t i nte r c ept . Thi s w a s improve d by a dding a fixe d e ff ect – condition – tha t i den tifi ed if the ta r g et or di st r ac tor wa s being cl assi fied ( along side a co rre spo nding r a n dom ef fec t f o r the pe r-pa rtic ipant slop e; Barr , Levy , She epe r s , & T ily, 2 013; A IC -874 6). Th e model w as further imp rov ed by adding a fix ed ef f ec t fo r N2pc ampl itude (and cor r e s pond i ng random e ff ect; AIC -702 ) and th e i nt e r a ctio n of t he se fac t ors (an d co r re spond ing r a ndom e f fec t: AI C -896 ). (eq 1 . ) cla s sifica t io n _acc uracy ~ N2pc * c o ndition + (N2pc * c on dition| p art i cip an t) ANOVA an a lysi s o f thi s mod el ide nti fi ed an inter action o f condition and N2pc (F(1 , 4 2.6) = 1 3.72 , p < 0.00 1), r efl ecting a signi f i can t di ff erenc e in t he s l ope o f th e l inea r fi t s illus trated in Fi gures 4b an d 4 c. Fol low-up mo de lling exa mined th e r el a ti ons h ip b etwe e n N2pc a nd ca t e gory in for mation s epa r a te ly for e ac h of the c ondi tion s. (eq 2 . ) cla s sifica t io n _acc uracy ~ N2pc + ( N2pc | partici p ant ) Targe t-e licit ed N 2p c r elia bly predi c ted i mprov ed targ et c la s s ific ation (F ( 1 , 41.6) = 5.366, p = 0.0 26 ), but di stra ctor -e lici t e d N 2 pc reli ably pre di cted de gra ded t arge t clas si fica tion (F (1, 42.0 ) = 6.13 0, p = 0.01 7). N o pat te r n was id en t i fi ed in se pa rate a naly s i s o f di st r a cto r- elici te d N 2 pc a nd dis t rac t or cl assi fic atio n ac curac y (F s<1 ). .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint Fi g ure 5 – Res ul ts fro m MRI cla s sifi c a t i o n. a.) S ea r chl igh t an a l ysi s i den tif y i n g t he loca ti on o f ta r ge t an d dis tract o r i nfor m at io n . T a rg et a n d dis tract o r c a te go r y i n f orm atio n e m erg e wi t h p art icul ar s treng t h in t he asp ects of occip it a l a n d pos te r ior t emp oral lo be s a ss oc i ate d w it h t he ve n tr al v i su al st ream . b.) V o x els iden tifi ed here ref le c t cl us ters ex pressi n g m or e i nfo r mat ion abo u t t h e target t han abo u t t he d istract or. I n b ot h pa n e ls t he lar g es t c lust ers e me r ge i n ve n tral v i s ual c ortex; o ther clu ster locat io ns are i d e n tified. OFC – orbi t ofro n t al cort ex; I P L – i n feri or parie ta l l obe; S PL – s up erior p arie t al l o be ; IFG – i n feri or fro n t al gyru s; T P J – tem poral pariet a l j unct io n. T he lef t h e mi s ph ere i s re prese n t ed on t h e l eft here a n d i n s u b sequ ent f ig ures . .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 12 Res ult s from E xperimen t 1 d emon stra te t h at the qual i t y of a tt entio nal s el ec tion o f t he targe t, refl ec t e d in targ et -eli cit ed N2pc , p redic ts an increa se i n t he ta rget c a tego ry infor matio n carri ed b y EEG in the same 125 ms t i me pe riod. I n contra s t , mi salloc a tion o f at ten tion to the distrac t o r , refl ec t e d in dis trac tor -elic i ted N2 pc, re su lted i n degr ade d t arge t ca t e go ry inform a tion. Measuring ca t e gory infor mat ion in uni mod al fMR I In E xperiment 2 par ticipa n ts c omplet ed t he s am e ta s k whi le EEG a nd fMR I were c onc ur ren tly record ed (Fig. 3b ). W e beg an an a lysi s wit h ide nt i fi cati on o f ca teg or y in forma tion i n t h e unimoda l fMRI signal . We es t i mat ed single -tr ial fM RI r e spon s e s us i n g the GLM singl e toolb o x (P r in ce, Ch are s t, Kurz a wski, P yle s, Tarr, & Kay , 2022) a nd i denti fie d ca t e gory in forma tion in brai n s pac e usi ng spa tial se archli ght s a nd L DA c la s s ifi ca tion . I n thi s analy ti c app r oa ch , a sph ere is d efi ned a r ou nd eac h vox el and clas si fica t i o n i s conduc te d ba sed on t he pa tte rn o f ac t i vity ob serve d in the su bse t o f vox els wi t hi n thi s volume (Kr i eg e skort e, G oebb el, & Ba nde ttini, 200 6 ). Se a r c hligh t sph er es ha d a diame t e r of 7 v oxel s (14 mm) an d c las si fie r s w ere t r a ined a nd te sted u s in g 4- fold c ro ss -valid ation . Thi s iden t i fi ed an ov erlappi ng se t of b rain a re a s t hat c on tain ed info rmati on abo u t both t h e ta rget a nd the di st racto r, alo ng s i d e a mo r e limit ed s et o f are a s tha t e xclu s iv ely cont aine d in f ormation a b out one of t h e s e t w o objec t s (F ig . 5a ) . Brain area s e xpr e s si ng categ or y in forma tion f o r both t arge t a nd dist r a ct or incl uded v en tr a l vi sual cor tex; in f e r i or a nd sup erior p arie tal lobul e s (I PL /SPL ) ; and pre - motor, mo tor, a nd soma t o s en sory co r t e x . Target i nfo r ma t i on al so eme r g ed in th e lef t inf erio r fro nt a l gy r us (I FG ), bi la ter al orbit of ront al c or t ex (O F C), a nd p r e cun eu s / cin gula te c or t e x; dist r a ct or informa tion ext ended fr om t he I PL into t he lef t t emporal -pa r i e t a l junc tion (T PJ ). Sta t i s tica l ana ly si s iden t i fi ed that the bil a te r a l LOC , IP L /SPL , lef t IFG , le ft OF C, po steri or c ingula te, an d le ft b a s a l gang lia (BG ) all re li ably contain e d more in fo rmat ion about the t arge t than di strac t or (F ig . 4b). N o a r e a relia bly c ont ained more info rmati on ab o ut t he di s t rac tor t h a n th e targ et. Coupling varianc e in N2p c to varianc e in MRI -derived category inf o rmation In E xperiment 1 we u sed a r e s a mpling ap proach to id enti fy within- par tici pant co v arianc e in inde pend ent me a sure s d erive d f r om EEG . In E xperiment 2 , we u sed a s imilar a pp roa ch to l ink va r i ance in mea s ur e s derive d fr om the t wo ima ging modalitie s (F igur e 2c ). For ea c h partic ipan t, t rial s we r e agai n s e p ara ted int o 20 chunk s o f e qual s iz e, w ith 17 c hunks as s i gn ed to the traini ng se t and 3 ch unks to the t e st se t. Th i s proc e ss o f sa mpli ng and model bu ilding was it era ted 114 0 times for e ac h of th e ~ 2.1 x 10 5 sea rchli ght s ph e r e s . In e ac h of th e se it era tion s, t he N2pc wa s me asu red in t h e te st se t a s mean v olt age acro s s a 125 m s in te rva l c entred at t he pe a k (a s iden tifi e d f r o m data c ollap sed ac r o ss i ter atio ns , condi t i o ns, and par tic ip ant s ; 189 – 315 ms ). Thi s ge ne r a te d a pa i red set o f 1140 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint Fi g ure 6 - Lin k i n g N2 pc a n d MRI ca te gory i nfo rma ti o n in E x perim e n t 2. R es ul ts i l l ustra te b rain r egi ons w here targ e t c ate g o r y i n for ma ti o n is p red icte d b y t arget -e lici te d N 2 pc a mpli t u d e. Ti me -c o urs e i nse ts s ho w t h e tim e - resol ved correlati on f or ea c h M RI cl u s t er. Si gnificant t im e cl ust ers are in dicat e d b y bl ac k li nes b el ow t he x- a x is an d s ha din g r e fl ects b o ots tra ppe d SE M. L o c atio n an d t im e-c o urse for a.) th e Inf erior Pariet a l L o bule clu s ter, b.) th e Ins ul a / P u ta m e n / Glo b us P al lid us cl ust e r, a n d c.) t h e V e n tro me d i al Prefro n t a l C ort ex / O rbi t ofro n tal Cortex clu ster. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 14 N2pc and c las s i fic a tion ac curac y v alue s f or eac h sea rchli ght s p he re. We ca lcul a te d the F i s he r- tran sforme d Pe ar s o n cor rela tion for e ach o f t h e se set s , ins er t i ng the r e sulting v al ues in to a new brain volume at the loc a tion o f ea ch se ar ch light c entre . A s i llu stra ted i n Fig ur e 6, t his iden ti fie d 3 cl us t er s wher e targ e t-el ici ted N 2pc ampl i tude pr edic te d MRI -de rived t a r g et ca teg or y informa tion . This inc luded a c lu ster in the le ft IPL / an gula r g yrus ext endin g into the SPL (92 vox els; Fig. 6a ), a cl us t er th at spann ed t h e putam en, globu s pall idu s, in sul a, a nd an te rior t empor al l obe in the r i ght hemi s ph er e (8 4 vox els, Fig . 6b), an d a c lus t er th at spann e d th e bila te r a l ven trom e dia l pre fron tal co r tex and orbi to fron tal c or t e x ( VMP FC / OFC; 16 3 vox els , Fig . 6c). We int erpr et t he se r e s ul ts a s demo n stra ting that i nc rea singly robu st a tten tion al s elec t i on of t he targe t – r e fle c ted in t a r g e t- elici te d N 2 pc ov er occ ipital c or t e x – cau sed the em erg enc e of c onc eptu al informa tion in t he ide n t i fi ed bra in a r e a s . H ow ever, a s i n Exp er i ment 1, the a l ter n ative i s tha t N2pc and categ o r y informa tion br oad ly c ovari ed as a func tion o f oth er varia bl e s a s s oc i ated to par tic ipan t sta te , like arou s al. W e ag ain c onduc te d a c ontr ol analy s i s to te st t h i s , rel ating t arg et c at egory informa tion to di s tr ac t or -e lic ited r ath er t han ta r g e t-e licit ed N 2p c . We c r e at ed 3 r egi ons o f int er est (RO Is) th at de s c r i be d th e c lu s te r s illus tra t e d in F igure 6, c alc ul ate d c las s i f i ca tion a cc ur a cy us in g the entir e s et o f vo xels wi t hi n eac h o f the s e ROI s , a nd u s ed our 114 0 -f o l d techni q ue t o relat e t hi s informa tion to t he N 2 pc. Th i s sh ow ed th at th e rela tio n s hip b etwe en t a r g et cat eg or y informa tion a nd targe t- elic it ed N2pc was signi fic antly gre ater than t he re la t i o nshi p be twee n t arge t ca t e go ry informa tion and di s tr a c tor -elici t e d N 2 pc in all 3 RO I s ( IPL / SPL: p = 0.005 ; ins ula: p = 0.027; V M P F C / OFC: p < 0.001) . D i st ract or -elic it ed N2pc i n fact pr edi ct ed a r e d uction i n MR I- deriv ed targ e t cat ego r y informa tion in t he V MPF C / OFC, much a s e me r g ed in analy s i s of uni m odal E EG r e sult s fr o m Ex periment 1 (p = 0 .019 ; oth er a rea s p s > 0.230 ). We c onduct ed whole -brai n analy s e s to r ela te dis t rac t o r-el ici ted N2 pc to varia nc e i n target o r dist r a ct or inf orma tion, bu t fa il ed to id en tify any se t o f vox els whe re th e se r ela t i o n s hip s survive d sta t i s t ic al c or r ec t i on . Analy s i s of di s tract or- elic ite d N2pc a nd di st r a cto r info rmati on ba sed on ROI s defin e d f r o m r e s ul ts illu stra ted in F igur e 6 also faile d t o iden ti fy sig nific an t re sult s . The a b s e nce o f relia bl e ef fec ts inv olvin g dist r a cto r info r mation may in pa r t r ef le ct t he limi te d magni t u de an d sc op e of thi s in forma tion i n th e MR I data (Fig. 5 ). Coupling varianc e in la t e r al EEG to varia nc e in fMRI-derived category informa t io n: time-c ours e To thi s poi nt , we ha ve foc u ssed o n the N 2pc me asu r ed a t a fix ed la t e ncy. H ow eve r, l at erali ty i n the EEG e merge s bo th be for e and a fte r t h e N 2 pc; a s di sc uss ed ab ove, e arly la te r a li ty emerge s a s a func t i o n of s e n sory activ it y , but al so a s a product o f stimulu s nove l ty (V erleg e r , ve l Grejew s ka , & .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint Fi g ure 7 – L i n k ing N 2pc a n d M RI vox el act iv at i o n in Exp e r i m ent 2. Res ult s i llust rate brai n r e gi o n s w h ere ta r g e t- elicited N 2pc a m pl i t u de is pre di c t ed b y vox e l acti v i ty. a.) Locat ion of t he t hre e i d e n tifi e d clus ters: Left Inf erior Parie t al L o b u le / An g u lar Gy r us / T emp oro parieta l J u ncti on , Left M i d dl e Fro n t al G y r us / Fro ntal Eye Fi e l d, an d Bil a tera l Dors ome d i al S u p eri or F ro n tal G yrus. b .) Time-res olved corr e la tion f or eac h MRI cl ust er. Sign ifica nt tim e c lus t e rs a re in dicat ed b y bl a c k l ines b e l o w t he x -a x i s an d s hadi ng reflect s bo ots tr app ed SEM. Ja s k owski, 2 012) and di s tr ac t or supp re ssi on ( Hic key , Di Lol lo, & McD onald, 2009; S awa ki & Luc k, 2010; Wea ver , va n Zoes t, & H i cke y, 201 9). La ter l at er a li ty ha s be e n cl osely link ed to the e ncoding and ma inten anc e o f s t i muli in visu al w ork ing me mor y (Vogel & Mac hizawa, 2004 ) . We e xpand e d our analy s i s t o ex plore the ro le o f pre - and p os t-N 2p c EE G l at er a li ty in t h e con trol o f c once ptu al enco ding in t he br ain. F o r each o f th e RO Is g en era ted from r e sults il lu s t rat ed in Fi gure 6, we r e la ted varian ce in eac h la ter al E EG sa mple t o va r i ance in MRI - d erive d t arge t ca t e go ry inf ormation . This genera ted a time - co urse o f th e r el ation shi p betwe e n lat eral occ ipit al EE G a nd M RI- d erive d ta r g e t categ or y inf orma tion f or ea ch R OI . The s e a r e illu stra te d in F igure 6. In the I P L , th e rel a t i on s hi p betwe en EEG later ali t y and t arge t c atego ry inform atio n emerge d very quick ly a nd su s tain e d (Fig . 6a) . Th e pre - N2 pc rela t io nship may refl e ct di strac to r s u ppre s s i on th at guides sel ec tive pr oc e ssing of th e comp l ete objec t ra ther t ha n t he deg ra ded obje ct in the contrala te ral visu al he mifi eld ( Wea ve r, va n Zoes t, & H ic key , 2017; H i cke y, Bert az olli, P o llici no , & Barbar o, 201 9) . The po st - N 2 pc r ela tion s hip, in con t ra s t, i s like ly to in dex th e m ai ntenanc e of v i s ua l working memory repre se nta tion s o f th e targe t i n po s t erior p a r ie t a l cor tex (Luria, Ba laban , Awh, & Vogel, 2018 ). In con tra st, i n th e VM P F C a nd insula c lu s t e rs, the r ela t i on s hip be twe en EEG lat erali ty and MRI - de riv ed ta rget in forma tion i s rel a tively discr et e to t he la t e ncy inte rval of t he N 2 pc ( Fi gs. 6b, 6c). T he s e r e s ul ts sugge st t hat the speci fi c selec t i ve proc e ss es in dex ed in N2pc ar e c los e ly li nked t o the pro paga t i on of targe t c at egory in f o r mation t o t h e s e are as . .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 16 Coupling varianc e in EEG indice s of attentional selec tion to varianc e in fMRI vo xel ac tivity In the t e xt abov e we have id enti fied a rel ation ship b e t w ee n N2pc ampli t ud e and t he emergenc e of MRI -derive d ca teg o ry informa tion . This a naly s i s w a s de s i gned t o i den tify th e e f f e ct of N2pc on the propag at ion o f c atego ry in f o r ma t i on d eri ved fr om visu al s timuli. H ow eve r, th e dat a also of fe r an opport unity to i d enti fy how N2pc i s pr ed ic ted by univa r i at e vox el ac tivity. Our analy s i s of the r el a tion s hi p be twee n univa r i at e fMR I and N 2pc wa s c once p t u a lly s imi lar to ou r ana lysi s o f multiva r i a t e fMR I and N2pc . C hunke d per -pa rticipa n t r e s ul t s were e xhaus tive ly parti tioned in to s ub se t s of 3 an d 17 . In e ach of th e re s u l t i ng 114 0 te s t se t s , th e N2 pc w as mea sur ed in a 125 ms w indow around it s pe ak and mea n ac t i vity w as cal cul a t e d fo r each o f the ~2 .1 x 10 5 vox els in c orte x and s ubc o rt e x. Th i s g ene rate d pa ire d se ts o f 11 40 N2pc and ac tiv ation value s for eac h vo xel. For eac h p artici p ant, we cal c ulated t h e Fi she r- tra n s fo rmed P e ar s o n c orr e l ation for ea ch of th e se pai re d s e ts , in ser ting the v alu es into a new br a in vo lume at e ach v oxel lo ca t i on. As illu s tr a t e d in Fi gur e 7a, t h i s i de nti fied 3 voxe l clu s ter s whe re v arianc e in targe t - eli cited N2pc ampl itude p redic te d varia nce in vox el ac t ivation : th e middle fr on ta l gyrus (MF G; 3 77 v oxel s ), t he I PL ex t e nding i nt o t h e TP J (486 voxe l s ), an d t he dor somedi a l sup er i or front al gyrus (d mSFG; 29 6 vox els ). The se a r e a s are la rgel y inde p enden t of t hos e ide n t i f i ed in c orre spondi ng ana ly si s of categ o r y informa tion . Signi f i can t re sul t s emerg ed i n the I PL in both a naly se s , but th e c lu st e rs did n o t overla p, wi t h N 2pc predic t ing ac tivation in vox el s ante rior and la te ral to tho se id en t i fi ed in a nalysi s o f ca tegory in forma tion . As in an a lysi s o f cat eg or y info rmatio n, w e ex amine d th e rela tion ship b e t w e en di stract or -elic it ed N2pc and vox el a ctivi t y as a c on tr ol . Who le-b rain a naly si s did no t i denti fy an y brai n a r e a wh ere th e dist r a ct or - e lici te d N 2 pc pr edic te d vox el a ctivi t y aft er s ta ti stica l co rrec tion . W h en we defin ed RO Is bas ed on t h e re s ul ts fr om whol e-b rain a na lysi s illu st r a te d in Fi gure 7, t he re was no signi fi cant relati on ship b e tween di stract or -elic it ed N2pc and mea n vox el a ctivity in the s e ar ea s, a nd th e relati on ship be tween ta r g e t - el icit ed N 2p c and voxe l ac t i vity w as s ig ni fica ntly gre a t e r t ha n th e relati on ship b e tween di stract or -elic it ed N2pc and v oxel activi t y (I PL: p = 0. 007; d mSFG: p < 0 .001; MFG: p = 0 .007). Coupling varianc e in EEG indice s of attentional selec tion to varianc e in fMRI v oxel ac tivity: time- cou r se As in an a lysi s o f cat eg or y info rmatio n, w e ex panded ou r analy s i s to t e st th e powe r o f pre - an d po st - N2pc EEG late rality to predic t uni varia te v oxel activ ity. As illu s tra ted in F igur e 7b, a rela tion ship betwe e n lat eral EEG and voxe l ac t i vity e m er g e s soone r i n IPL th an in MF G, t ho ug h both e ff ec t s a re in the 17 5 - 350 m s in t e rv al wh er e t he N2p c c an e merge. Bo th l e ft I PL a nd le ft MFG hav e bee n clo sely .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 17 link ed to th e e xtrac tion o f sema n t i c c ont ent during r ead ing ( Wu, Ho, & Ch en, 201 2; Rue ckl et al ., 2015 ; Binder e t a l. , 2009 ), and l e f t a nd ri ght IPL and MF G have r ol e s in the o rien t ation and reori enta tion o f spa tia l at ten tion (Co rbe t ta & Sc hulman , 2 002; Co rbe tta , P a tel, & Sch ulman, 2 008). One int erpr et ation o f th is finding i s tha t the eme rge nce o f I PL a nd MFG r efl ec t co ntrol in th e depl oyment o f a t ten tion for e xt racti on o f sema ntic in forma tio n . The r e la tion ship be tw een la t e ral EEG an d activ ity in the dmSFG , in contr as t , em e rges from very early , inc luding lat enc ie s t h a t pre cede sti mulus on se t. Amo ng othe r func tion s, thi s por t i on o f medi al fron tal lob e contai n s e ye fie l d s a nd ha s b een a ssoc ia ted w ith fixation moni toring, i ncludi ng during smoo th pur s ui t e ye mov ement s (Amiez & P e t rid e s , 2 009) . In th e curren t re s ul t s, trial s with e ye move ment s sub stan tivel y prece d ing s timulus on s et wer e n ot exc luded from a n alysi s; t he a s s oc ia tion of ac tivity in th e dmSFG with E EG la te rali ty underlying the N2pc may refl ec t ocul o motor cont rol suppo rt i ng ori ent ation of th e eye s t o the fi xation m ark or in ma in ten anc e o f the e y es a t tha t loc ation . Similarly , lat er a s socia tion o f d mS F G w ith EEG la te r a li ty in t he N 2 pc a nd post- N2pc inte rval may reflec t the involv eme nt o f thi s brai n a r e a in th e inhibiti on of s a cca d ic mo vement s to co ve rtly att ended loca t io n s. Dis c us sion We le ver aged nat ural va r i a bility in the q ua lity of a t t entio n al selec t i on t o identi fy t he t emporal a nd ana t o mic c har acte ri stic s of the a t t entio n -me dia ted pr opag a tion o f c oncep t u a l inf ormation i n the huma n brain. In Experime nt 1, we find th at th e a mplitud e of targe t - el icit ed N 2pc , an index of s pa t i al att entio n, p r ed i ct s t ha t t h e evok ed EE G s igna l wi ll ca r ry more info rmation a bout t he s emantic ca tegory o f a n at tend ed o bject (Fi gure 4 b). T hi s r e la tion ship emerge s qu ickl y, w ith atte ntion - medi ate d ca t e go ry informa tion emergin g in the sam e 173 - 298 ms window u sed in me as u reme n t o f the N2pc . The mi s a l loca t i o n of a tt enti on to ta sk -irr eleva n t stimul i, re flec t e d in di s tr a ct or -elici te d N2pc, pr e dic t s a reduc t i on in th e quali ty of t a r g et ca teg ory in forma tion i n th e sa me interv al (Fig ure 4c ). Though the se r e s ul t s are c o r rel a tion al, the p att er n s t rongly s u gge s ts tha t a tt ention a l s elec t i on of the t arget rapi dly c au se s eme rgenc e o f c ategory i n f o r ma t io n in bra in activ ity. Ex periment 1 l eave s unc le ar th e degr ee t o w hich t he ob ser v ed rela tion s hip be twe en N 2pc and ca tegory in forma tio n r e fl ect s low-l e vel p er c ep tual encodi ng. Tha t i s , ou r machi ne lea r ni ng app roa ch to the d e r iv ati on of c a tego ry informa tion co uld rely on the r epr es ent ation of low - and mi d-level v is u a l fe at u r es t h at ar e s ha re d b y e x a m p le s of a g iv e n ca te g o r y , a n d t h is p oss i b il ity i s n o t e nt i re l y addre s sed by our u se of a vi s ual ly he ter o ge nous stimuli se t. For exa mple , a tten tio na l s elec t i o n of an ima ge of a vehic le ma y c au se r epre se nta t io n of the surfa c e c harac te r i s tic ‘metalli c ’ more of ten tha n .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 18 is th e c a s e f or any o t h e r i mage ca tego r y i n the e xperim ent s. Thi s wo uld provid e o pportuni ty fo r the cl assi fier to le ver age th e emerge nce o f b r a in ac tivity a ssocia t e d with t he pe rcepti on of thi s vi sual fea tur e to pr obab ili st i c all y infe r targ et c a tegory. Str o nge r at ten tional acc e ntua tio n of thi s br ain ac tivity c ould i n thi s way improve catego ry cla ssifica t i on . One goa l o f Ex periment 2 w a s th ere fo r e t o cla r i fy the co ntributi on o f visual p er c e ption i n th e relati on ship b e tween a tten t i o n and c at e gory i nfor m a tion. Initia l anal y s i s o f unim odal fM R I iden tifi ed a n etwork o f brai n a rea s whe re cat ego r y informa tion w a s grea te r fo r t a r g et s th an for di s t rac tor s (Fig 4b). T hi s pr om ine ntly i nclude d bila te ral ventr al v is u al c ortex , alon g s i d e small er c lu s t er s in bi la ter al inferi or pa rie ta l lobule / sulc us (I PL/SPL ), lef t inf erior fron tal gy r u s (I F G), l e ft o r bi t o fron tal cor tex (OF C), l ef t ba sal gang lia, a nd bila teral p o s t erio r c ingula t e gyru s ( PC ). The se re sul t s t he r e f o r e inclu de asp ec ts o f oc cipi t a l and p oste rior pa rie ta l cortex tha t a r e strongly a s s oc iat ed w ith the re pre s enta tion of v is u a l fe atu re s and t ran sla tion o f v is u a l informa tion to mot o r c on trol (Konen & Kastn er, 2008 ; Freud, Plaut , & B ehrmann , 2016), a long si de ba s al gang lia struc tur es tha t m ay ha ve a r o le in v i sual perce p tion ( Ding, 2 023) . Howev er, al s o e v ident are a se t o f le ft - l a ter alized a rea s i nc luding IFG and VMP FC / OFC , a long side b il ate ral P C, th a t are no t put ativ ely v is u al in n ature , bu t hav e been impli cated in ea rlie r w ork as c on taining a bstrac t s em a ntic re pre s enta tion s ( Binde r et a l., 2009 ) . Furthe r ana ly si s iden tifi ed 3 brain a rea s w here tri al-wi se v ari anc e in t arge t - el icite d N 2pc ampl itude predic te d th e qua lity o f ta rget c a tegory i nfor m a tion : the l e ft I PL /SPL ; a n ins ula -c e ntr e d cl u ster in t he right he mi s phe r e incl uding po rtion s o f th e puta men and globu s pall idu s ; a nd a bil ater al VM PF C- ce ntred clus ter extendi ng int o me dial O F C (F igur e 5). The s e are al l node s in t h e ‘ se mantic n e t w ork’ iden t i fi ed from larg e- sca l e met a- an alysi s of f MRI s tu di e s o f sema ntic r epre s ent ati on (Bind e r e t al., 2009 ). The po rtion o f I PL/SPL identi fie d i n this a na lysi s w a s phy sica lly clos e to th e pos terio r pari etal cl us t er ide n t i f i ed in a n alysi s o f unimo dal fMRI , a s de ta iled a b ove, b ut wa s s l ightly posteri or wi t h no ove r l ap, em e r g ing primaril y in th e angula r gyrus. Cat egory inf ormati on in thi s ar ea expres se d a relati on ship w ith ERP l ate rality o ver muc h of 500 m s f oll owing stimulu s on se t, inc luding time perio ds befo re a nd a f ter N 2 pc. Ac tivity i n thi s po r tion o f th e parie tal lob e i s imp lica t e d in the c ont r o l o f vi s ual at ten tion (Bi sley & Goldb erg, 2010 ; Ca b eza, Ciaram e lli, & M o scovi t c h, 201 2 ) but al so in t he retri eval and r epre s ent ation o f e pi s o di c ( Se stie ri, S hulman, & Co rbe tta , 2 017; Wa gn er , Shann o n, Khan , & B uckne r , 2005) an d s e ma ntic i nf or ma t i on (Bind er e t a l., 200 9 ). In lin e wit h other rec en t theory ( Humphrey s, L ambon Ral ph, & Si m ons , 2021) , our t ent ativ e ac coun t i s tha t thi s a spec t of IPL / SPL is a hu b in the compa r i son o f v isu al c haract eri stic s o f th e at tend ed stimul us t o c atego ry templa t e s re tri eved from memory, i n thi s way acting a s an imp or t ant w ay sta tion in t he at ten tion - medi ate d cl assi fic a tion and ab s t rac tion o f vi s ua l inp ut. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 19 Inte rpre ta tion o f the c lu s t e r cen tr ed on t he r i ght in sula i s fu nda men tally chal lengi n g bec aus e of the bread th of fun cti onal r ol e s a ssoci at ed with t hi s a rea (Udd in et al. , 2017; G a squoin e , 2014). A co mpelling pos s ibi li ty, how ever , i s that t his r efl ec ts th e role o f in sula a n d s u rroun ding brain ti s s ue in the ‘ sali e nce ne twork’, a s e t o f br a in a re as impl ica ted in th e id en tifica tion o f hom eo sta tic rel e vanc e and behav iour al impo r t anc e (Se el ey e t al ., 2 007). The s a li enc e ne twork i s thou gh t t o guide t h e flow of inf orma tion t h roug h oth er br ain ne t w or k s in t he suppor t o f at ten tional proc e ssing and o ther co gnitive oper ation s (Men on & U d din , 2 010 ) . U nl ike I PL/SPL , ca teg ory info rmatio n i n insula co r rela t e d di scr et ely with ERP lat erali ty in the la te ncy of the N2pc, i mplica t i ng th e N2pc s p e cific ally in the pr o p agatio n of in forma tion to thi s area . The eme r g ence o f at ten tion -media ted cat eg or y informa tion in t his clu s ter s ug ge sts a rol e for th e in sula a nd surroundi ng ti ssu e i n t he proce ssing o f abs tr a c ted ob jec t info rmatio n t o guide e valu ation o f beh aviou ral r e leva nce a nd t he prepa ra tion o f a co gnitive r e s pon s e, po s sibly in the c on te xt of c ur ren t pa rtic ipan t s ta te and ta sk motivation . The clust er cen tre d on VMP F C is the la r g est of the 3, ha s the st r on ge st rel a tion s hi p with N2pc , and uniqu ely sh ows a link t o b ot h ta r g e t-e lici t e d and di stract or -elic it ed N2pc, with tar get c atego ry informa tion red ucing in thi s ar ea a s a f un ction o f t he mi s-d eployme n t of a tt enti on to the di s tract or. As with the i n s ula , thi s cor tica l are a h a s b een a s socia ted with ex ten s i ve fu nc t i on ality (St alnak er , Cooc h, & Sch oenb aum, 2015). Promin e n t is the ide a t ha t it car rie s inf orma tion a b out th e ab stra cted ec onomic utility o f g ood s and a cti on s - t he ‘c ommon c ur renc y’ hy po the si s (L evy & Gli mche r, 20 12; Pad oa -Sch iop pa & Cone n, 2017). T hi s kin d of s ign al is n ec e ssary in ec on omica lly n or ma t i ve mod el s of le a r n ing an d dec i s ion maki ng to all ow the u t il i ty of di spa rat e g ood s and ou tco mes to be dir ec t ly co mpared. H owe ver, ou r re sul ts sh ow th e emergenc e of a t t enti on-m ed iat ed c at e gory in f o r ma t i on in this are a w hen c atego rie s do no t e xplic itl y di ff e r i n val ue. Thi s i s in line w ith th e kn own inv olve ment of thi s a rea in s e man tic s (Bind e r e t al. , 20 09), but i s dif fic ul t to re solv e w it h th e n otion th at t he c or e re spon s ib ili ty of thi s br ain ar ea i s to r epr ese nt valu e ab strac ted from c on t e xt and stimuli . Rece n t model s o f OF C / VM PF C functi on s ugg est that th is b rain reg ion ma y be b etter d e sc r ibe d a s invo lved in the de rivation o f ta s k s t ate f rom d i s pa r a t e attribu te s of en vironmen tal objec ts and contex t (Pe ll eti er & F ellow s , 2 021; Kle in -Flüg ge, Bong ioanni , & Ru shw orth, 2 022) and in t he e stabli shm ent o f co gnitive maps o f t ask spac e (H a yden & Niv, 2 021; Niv, 2019; Schuc k, Ca i, Wil s on, & N i v, 201 6). The cu r ren t re sul ts a r e e a sily des c ribed in thi s th eor etic al c on text: s tr ong at ten tion al sel e ction o f a beha viourall y -re leva nt vi sual ob jec t le ad s t o t he ac tiv ation o f a cog nitive ma p in OFC / VM PFC t h a t inc ludes in fo rmation a bout the targe t ob je ct ca t e gory an d i t s a s s ocia te d re s po nse . W he n a tt entio n is mis- deploy ed to th e di str actor , ta rget in f ormation in thi s cog nitive ma p i s degrad e d. As no ted a bov e, a motiva tion for E xperi ment 2 was t o addr e s s t he po s s ibi li ty tha t the r ela t i on s hi p betwe e n N2pc and c at egory in forma t i o n identi fied in r e s u lts from Ex perimen t 1 w as drive n by the .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 20 enc oding of vi sual f e a tur e s . In t h i s c on te xt, one dat a pa tter n i n Expe r i ment 2 i s n otable for i ts abs enc e. We do n o t see t ha t t h e N2pc r o bus tly predic t s MR I -de r iv ed c a tego ry i nfo r ma t i on in v i s ua l co r tex . Thi s null findi ng is , o f cour s e, am bi guous, bu t sugge st s th at ou r e xperime nta l de sign w a s succ e ss ful in ident ifying inf ormati on a b s t racted fr om the visu al f e atu r e s th at c har acteri s ed the st i mulu s. That i s, ou r s timulu s se t wa s vis uall y hete roge nou s, with few low - or mi d -level vis ual fea tur es tha t coul d be c on s i s t ently u s ed t o inf er ca te gory memb ership . As a res ul t, c atego riza tion relied on a s et o f idi o s y ncratic di agno stic fea ture s for e a ch i mage - for e xampl e , cl assi fic ati on of a cu pcak e as food r eli ed o n a dif fe rent set of fe at ur e s t han did c la ssific a tion o f a co urge t t e a s f ood. Bec au se a tt ention ac ted o n a varyi ng se t of rel ev ant fea tur e s fo r e ach image , a tt e ntional sel ec tion did not cau se c on s i s t ent ac c en t ua t i on o f a pa r tic ula r s et o f f ea ture s , and th ere fo r e di d not pro duce a co ns i st ent p at tern o f ac tivi ty i n visual cor tex that c oul d be leve ra ged by the c la ssi f ier to i mpr ov e perf ormanc e . In t his , re sul t s fr om Ex periment 2 pr ov ide some r ea ssura nce t h a t fi ndings f rom unimod al EE G a naly si s in E xperiment 1 ar e not drive n entir ely by the re pre s enta ti on of v isu al fea tur es . Import antly, thi s d o e s n ot imply t ha t t h e att entio n-me diat ed r e p r e s e nta tion s th at w e identify in pos ter i or pari et al c ortex , insul a, and pr ef rontal c o rt e x ar e ne ce s s a r i ly amo dal or i nst anti ate d inde pend ent o f sen so r i moto r sy st em s ( M a hon & Car amazz a , 2009). The r ela tio n s hip be tween N 2pc and categ o r y informa tion w e ide n t i fy is n ot a s imple r e fl ec tion o f at t e n t i onal ef fec ts o n visual fea tur es , but i s lik ely to c on tain sema n t i c fea ture s that a re g r ou nd ed in sen so rimo tor sys tem s, inc luding informa tion ab o ut t h e s pe ci fic motor re spon se a ssocia t e d with e ach cat egory (Bar sal ou, 2020 ). The pa t te rn w e ide n t i f y - tha t at te ntional e f fect s on p erc ep tual r epr es ent at ions in la ter al oc cipital cort ex pr e dict s t h e quali t y of pe rce ption -a bs t rac ted in forma tion e lsew he re in t he b rain - is ve r y much in line with mod er n acc ount s of ground ed s eman t ic repr e sen tati on ( C hen, Lambon Ralph , & Roger s, 2017 ; M a rtin, 2016) . In a dditi on to ou r ide nti fica tion o f at ten ti on-medi at ed pr op aga tion o f s ema n tic in fo r ma t i on , re s ul ts from Ex perimen t 2 prov ided t he op po r tu nity to iden tify br a i n s t ruc ture s wh ere N2 pc am plitude predic te d unimoda l fMRI ac t i vatio n. Thi s identi fie s a se t of b rai n ar ea s inc luding le ft MF G (incl uding the le f t fron tal eye fie ld s; Balda uf & De si mone, 2 014), le f t IP L ex te nding i nto TP J, and bilat eral dmSF G (Figur e 7 ). All of the se a rea s ha ve doc umented role s i n a tten tion al c ontr ol. As no ted ab ov e, the F EF a nd I PL are co re nod es in the fro ntopari et al netwo rk imp lementing s t rat e gic atten tional co nt rol, and the b roa d er MF G and la t e ra l pos t e r io r pa r i etal c o r tex have be en impl ic ated in th e u s e of at ten tion for seman tic ex t r ac tion . Si milarly, dmSFG c ont ain s eye fi eld s and play s a rol e in s acc adic co nt rol . This s ug ge sts t hat a c tivity in the s e area s ma y be i nvolv ed in th e e s t abl i s h me nt and inst anti ation o f a tten t io n al c ontrol, s uc h that their ac tivation ca us e s an i ncre a se i n N2pc ampli t u de . .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 21 This a ccou nt i s r ei nfo r c ed by th e temp or al pa t t ern o f th e r e la tion ship b etwe en d mSF G and N2pc , and betwe en I PL a nd N2pc, wher e la te r a l ity in t he ERP pr ec eding t h e N 2 pc w indo w pr e dict s activ ity in the se a r e a s (F ig 7b ) . In c oncl us i on , we t r ac k the i mpact o f spa tia l at ten tion on the s pr ead o f conc eptu a l i nfor ma t i on in the huma n br ain. Our re sul ts de mon stra t e the tim e-c o ur s e o f thi s p ropa gati on of i nfor m a tion and iden t i fy thre e key brain st r u ctur e s c arryi ng abstrac te d, a tte ntion -medi a te d info r mation: the la ter al pos ter i or pari et al c ortex , the a n t e rio r i n sula , and t h e ventr omedia l pr e fron t a l co rt ex . The e mergenc e of la te r a l p os t e rior pa r i e t a l co r t ex in thi s con t e xt i s clos ely in li ne wi t h pr i o r re s ear ch on spa tial att entio n, b ut the iden ti fic atio n of in sula and VM PFC i s un expe cted a n d, a s a re sul t, co n s tr ain s and gui des f u nctio nal int erpre ta t i on o f th e se s t ructur e s in broa d er con tex ts. Thi s brai n network interve n e s b etw een p e r c ept ion a nd ac t i on, tr a ns l a ting v is u a l inf ormati on to c onc e pt u al kn owledg e and action pla ns .

Human participants All pa r tic ip ant s wer e r ec rui ted from t he Unive rsity of Bi r min gham c ommunity, pro vided writte n inform ed c ons ent , rep orte d normal o r c o r rec t ed -to -no r ma l vision , and rep ort ed n o h istory of neurolog ica l o r psychi a tr ic di sord er s . All demograph ic s col lect ed from e xperime n tal par ticipa n ts a r e repor ted an d r e sul ts ca n be ex pect ed t o general i se t o t h e Europ ean popul ation fr om w hich t h e sampl e was t aken . Expe r i m ental p roce du re s w ere app rov ed by t he Un iver s i t y o f B irmingh am ST EM ethic s c ommitt ee a nd the C HBH H e al th a nd Safe ty c ommitte e. Experiment 1 Forty- nine pa rticipa n t s w ere r ecrui ted. T wo w ere ex cluded f r om analy s is d ue t o e xc es s ive eye move ment ar te fact s in th e EEG ( r e jecti o n of >40% of tr i al s), t w o we re exc luded a s re sul t s s how ed no co ns i st ent evi dence o f a ta rget -eli ci ted N 2pc (ie. in t h e 1140 ite r a t i o ns e mploye d f or cla ssi fica t io n, the la t e ra l targ et -elic it ed signal s how ed c ontr a la te ral po sitive p olari ty more o ft e n than neg a tive pola r i t y ; p er -par t i cipa n t p < 0.0 5, bin omia l te st), and a sing le p artic ip ant wa s e xcl uded due to low ac curac y in t a s k p er for ma nce ( < 70% a cc ur a cy; f i nal sampl e of 44; a ge mea n = 21 . 8, st d = 2.7; 2 3 femal e; 3 le f t -han d ed) . The experim e nt too k approx imately 2.5 ho ur s, inc luding EE G prepa rati on, ta sk traini ng , break s, a n d debri ef , a nd pa rticip an ts w e re co mpen sa te d at £10 / h ou r. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 22 Experiment 2 Thirty-ni n e par ticip ant s w e re rec ruit ed. T wo pa r tic ipan t s were exc luded from an a l ysis du e to ex ces s i ve ey e mov emen t ar te fac ts in t he EEG (rejecti on of >40% of t r i al s), 4 partic i pant s were ex clude d due to i ncompl e te data c oll ec t i on due to tec hnic al i s s u e s o r e quipmen t fail ure , and 2 partic ipan t s were ex clude d due t o l ow ac curac y i n tas k pe rf ormance (<70% acc ura cy ; final s ampl e o f 31; age me an=23 . 2, s td =3 .8; 19 fe male ; 3 left -han ded ). T he ex pe riment took a pp roximatel y 4 h ours, inc luding EEG prepar ation , M RI s c re enin g, t a s k tr a ining, b r ea k s, and d e br i ef, and p art ic ipan t s we r e co mpens at ed at £15 /hour . Experimental task Par tic ipant s in bo th ex perim ent s pe rf orme d a v is ua l sea rch ta sk a nd a r epe t i t i on detect ion t a s k in interl ea ved bl ock s (4 bl ock s of 12 8 t r i al s for e ach t a s k with c ount erb alanc ed o rde r). Compl e tion o f the v is u a l s earch ta sk took longe r - app ro xima t e ly 75% of the e xper iment du rati on - and on ly thi s ta sk is t rea ted in t he curr ent p ape r. In e ach experim en t a l tr i al , pa rtic ipant s w ere a s ke d to rep ort t he s e ma ntic c at egor y of a targe t obj ect (Fig 1a) th at a pp ear ed in a c ued loc a tion an d to ignor e a di st r ac tor o bject tha t a pp eared i n a n un- cu ed loc ation (Fi g 1b). Each trial b egan with a fixati on di s p lay on white b ack groun d (1550 – 1850 ms , randomly s elec ted f r om a unif orm di strib ution) tha t co nt a in ed 4 grey c ircle s (E1: 5 .7° vis ual angle ; E2: 4.2° ) direc tly abov e and b elow (E 1 : 7. 4° vi s ua l a ngl e; E2: 5.2 °) a c entra l fixa t i o n mark. T he fix ati on mark was repl ac ed by the l et ter ‘ R’ or ‘ G’ (135 0 – 165 0 ms), whic h indi ca ted t he red or gre en c olor o f the c ircle id enti fying the t arge t in t he s u bse que n t sea r c h a rray. I n the se arch a rra y , imag es o f co mplete obje c t s appe ared withi n the tw o circ les that h ad red o r g r e en c ol or. The ob ject s were tool s, veh icle s, a nimal s, or food s, a nd we re s elec t e d fr o m an ima ge se t co nt a i ning equal n umber s o f eac h c atego ry (E1 : 160 ob ject s in to tal; E 2: 120 objects ; Fig . 1a) tak en from th e B OSS dat aba se (Brod eur, Dionn e- Do s t i e, M ontre uil, & Le page, 2010) o r ope n o nline image re po sitorie s . Ima ge ex amples w e r e s e lec ted from e ach c ateg ory r a ndomly without r epla cem ent un til the se t was ex haus t e d and the proc e s s re s et. Th e oth er t w o ci r c le s in th e d i s pl ay c ontai ned u nr e cogni s abl e morphed ver s io n s o f th e sam e imag e se t (Stojan oski & Cu sac k, 201 4; max dis t or ti on 16 0, ns t e p s 4 , itera tion 9 ), s el ec ted i n a simila r m ann er . The s ea r c h arr ay su s t a ined for 60 0 m s a nd the ne xt tr i al beg an eith er i mmedia tely a ft er re s p on se or a f t e r 1 800 ms . The expe r ime n tal ta sk w a s t o id enti fy, v ia righ t hand r e spon s e , whic h obje ct ca t e gory ap pear ed in the c ued c ircle . Two of t h e cat eg or i e s we re a s s o ciat ed with inde x fing er r e s p on se an d the o the rs wi t h middle fi nge r re s p on se, with th e ma pping o f cat eg or y to r espon se c oun terb a lan ced a cro ss partic ipan t s . I n Ex periment 1 r e spon se w as ma d e via a s tanda rd k eyboard ; i n Expe r i men t 2 re spon s e .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 23 wa s v ia a n MRI -compa t i ble bu tto n box . The r e spon s e a s s o ciat ed with th e ta rge t o bject w as alwa y s oppo site to th at a s socia t e d with the d i s t r actor obje c t. The expe r ime n t began with v erb al ins tr u c tion s and pa rtic ipan t s s u b seque n t l y c ompleted traini ng until the y ac hi eve d ~80% ac curac y, wh ich rarely to ok l onger th an a s in gle bl ock o f t ri a ls. In Ex periment 2 tr aini ng took p lace out s id e of th e sca nn er. Eac h expe r i men tal blo ck w as compo sed o f 128 t ria l s i n random o rder , w ith eac h t r i a l refl ec ting one combi n ation o f c ue c olo r, target and dist r a ct or cat eg ory, and ta r g e t and di s t ra ctor po siti on. In Experim e nt 1, st i muli we r e pre sen ted via a 59.5 cm x 33.5 cm L CD monitor a t 60 H z and a dis tance o f 60 c m. In Experimen t 2 , s timul i we re pre sen ted vi a a P roPi XX projec tor a t 1 00 Hz v iewe d via a mirror mount ed on th e sc anne r h e ad c oil. The expe r ime n tal proc e dur e relie d on th e P sychT oolbox 3 toolbox (Bra inar d, 199 7) for M a tl ab 2017 a. Data acquisition and preprocessing: Experiment 1 Ele ctrophy siologi cal da ta wa s a cquir ed fr om sint ere d Ag / Ag Cl elec t rod e s at 1 kHz using a Bio s emi Activ eTwo amplifie r . EEG wa s c ollec te d f r o m 128 e lectr ode s fi tte d in a n ela s tic c a p a t equidi s tant enc ephal ic sit e s, horizonta l elec t r ooc ulo gram (H EOG ) wa s c ollec te d from 2 ele ct ro de s locat ed 1 cm later al to t h e ex terna l ca n t hi o f the l ef t a n d r i ght eye , ver tic al el e ctro oculog r a m ( VEOG) w a s co llec t e d fr om 2 el ec tr o d e s i mmediat el y abov e and below th e right ey e, and unu se d re fere nce signal s w er e col lect ed f r om 2 elec trode s pla ced over th e le ft a nd right ma stoi d pr oc ess e s. EEG w a s re sampled o ffli ne t o 200 H z, dig itally fil te red wi t h s ym metr ic Hamming wind owed fin it e -impul s e re spon s e ke rnel s (h igh-pa ss a t 0.0 5 Hz , - 6 dB at 0.02 5 Hz ; lo w -pa s s a t 45 Hz , - 6 d B at 50. 6 Hz ) , ref ere nce d to the av e rage o f encep h alic cha nnel s, and b a sel ined o n t he 200 m s interva l pr ece ding st i mulu s on se t. N oi sy channe l s wer e vi s u a lly id enti fied a nd in ter pola ted u sing sph e r i cal splin e interp ola t i o n. In depe n den t compo nen t a nal ys i s ( IC A) o f c ombined EEG and E O G d ata wa s us ed to iden t i fy data v ari anc e re s u l t i ng f rom ey e movem ent s . T r i al s with ey e mov ement s i n the 500 ms interva l follow ing s timulu s on se t were r e mov ed from ana ly si s and t he I CA compo nen ts re fl ec t i ng ey e arte fac ts we r e s ub seq uen t ly remove d f rom t h e dat a. Data acquisition and preprocessing: Experiment 2 MRI Whole -bra in s c a nning employ ed a 3 T Sie mens Pr i s ma MR I s c an ner a n d 64 c hanne l head-c oi l. Fun ctional da ta w a s a cqui red u s in g an e c ho pl anar ima ging (EPI) se quence wit h 57 axi al slice s of 84 x 84 v oxel s , fi e ld- of -view (FO V ) 210 mm x 210 mm, slice thick ne ss 2.5 mm, sl ice ga p 0 mm, repeti tion time (T R) 1.5 s , ec ho tim e (TE) 35 ms , flip a ngle (F A ) 71°, mul ti-b and acc ele ra tion fa ctor 3. S t ruc t u ral data wa s ac quir ed u sing a T1-we ight ed MPRA GE s eque nc e wi t h 208 a xial s li ce s o f 257 x 257 vo x els, .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 24 FOV 256 mm x 2 56 mm, slic e thic knes s 1 mm, slic e gap 0 mm, T R 2 s, TE 2 .03 ms, FA 8 ° . Fo r co r rec t i on of o the r imag e s , a fi el d map w as acq ui r e d u sing a doubl e- ec ho sequ enc e with 3 6 ax ial slice s o f 6 4 x 6 4 vox els, FOV 192 m m x 192 mm, slic e thic kne s s 3 .75 mm, s lic e ga p 0 mm, TR 400 m s , TE 1 4 .92 ms, TE 2 7. 38 ms, FA 45 ° . Fo ur -le ad vectorc ard iogram ( VC G ) was r ec or d e d in the sc a nner from el ec tr o d e s i n a stan dard c he st m on t a ge. VC G w a s r ec or ded a t MRI cloc k s pe ed (5 kHz ) an d wa s synch r oni sed with thi s s i gnal. MRI ima ge s were co rr e c ted for he a d mo veme nt, f ield m ap inhomo ge neity , a nd sl ic e-acqu i s i tion dela y (u sing the middl e s lic e a s re fe renc e ) bef ore b ein g normali se d to M on tr eal Ne ur ol ogic al Ins titu t e (M N I) spac e, int erpol ate d t o 2m m isot r opi c v oxel s iz e, a nd s p a tially s mo o the d u s i ng an iso tr op ic G au s s i an ke r n e l (6mm full widt h ha lf max imum). G ene ral l inea r mode l a nal ysi s ( GLM) o f the f MRI time- s erie s d ata r eli ed o n the G L M sing le t oolbox ( Prince, Cha re st, Kur za w ks i , Pyl e, Tar r, & Kay , 2022 ) . G L M s in gle id en ti fie s hemody n amic kernel fu nc tion s, d erive s nui sanc e regre s s o rs , and ch oose s an optimal ridg e regul a r i sa tion shrinka ge fr ac t i on for eac h vox el to produ c e an e s tima te o f single -t rial voxe l-wi se b eta value s . T o ali g n EPI v olumes wi t h t he on se t o f visua l s timulatio n , th e EPI time s eri es wa s up - sample d to 6.7 Hz (15 0 ms per im age ) prior to li ne ar model ling using sha pe - pre serving piece wi se cu bic i n t e r p ol ati on (a s imple men ted in th e tse rie sin terp .m f unc t i on inc luded in knk utils to olbox; http s: / /gi thub. com /cvnl ab/knkutil s ). An al ys i s w a s con stra ined to v oxels out side o f the brai n stem, the ven tricle s , a n d the c e rebellum . EEG Ele ctrophy siologi cal da ta wa s a cquir ed in the sc an ne r fr om s in ter ed Ag /AgCl el ec trode s u sing a Brain Produc t s Brain Amp MR amplifi er . EE G was c ollec ted fr om 63 ele c trod e s f i t t ed in a Brai n Cap MR I- co mpatible el a st i c c ap a t ex tend ed 1 0/ 2 0 ence phalic sit e s a nd singl e-l e ad el e ctr o cardiog r a m (ECG ) wa s rec o rded from an e lect rode l oca ted o n the bac k 10 cm c auda l to th e top o f th e sho uld er a nd 5 cm t o the le ft o f th e s pine . All dat a acqui s iti on was a t MR I c lock s p e ed (5 k H z) and sync hroniz ed wi t h this s i gnal . Pre proc e s s ing bega n with c or r e ction for M R gradi en t ar tif act s u s i ng Br ainVi sion A na lyzer 2 (v 2.2). Gradi ent trigge rs sen t by the s c a nner we re used a s t i me ma rke rs f or ar tif act trai ns and co nt i nuou s ar t i fa ct s wer e c or r ec ted u sin g templ at e sub t rac tion ba s ed on a sliding averag e of 21 TR interva l s (w i th ba selin e corr ection ). The EEG w as s ub sequ e nt l y dow n s am pled to 250 H z an d co r rec te d f or cardio ba lli stic ( CB ) a rte fac t s . T he CB c orrec tio n alg orith m s e mi -au to matica lly det ect s pea ks fr om h ear tbea t signal s. In t he 124 e xperiment al block s analy z ed ( 4 for eac h of 31 partic ipan t s ) , t he s ig nal qu a lity a nd av ail abi lity of h e artbe a t s ig nal s varie d: V CG w as u sed for co r rec t i on in 52 block s , ECG wa s u sed in 48 bloc ks, and the r e m aining 24 blo cks r elie d on a h e ar t - bea t I CA compon e nt d erived f rom th e EEG. A run ning averag e o f 21 bea t s was u s e d to gen era te a templa t e and the a rti fac t tim e del ay w a s automa tic ally e s tima t e d u sing a 40 s in te rval. Sub sequ ent .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 25 dig it a l fil tering, b a selining , noi sy c hann el interp olati on, I CA, a nd tri al r ej ecti on p ara mete r s w ere a s desc ribed for Experime n t 1. The ave rag e number o f i nt e r pol a ted c hann el s wa s 0 .2 9 and t he numbe r of ind ep ende nt c ompone nt s rem oved was 2.0 . The mea n numb er o f c orr e c t tr i al epo ch s contai ni ng ex ces s i ve noi s e o r e ye mov emen t s was 2 7 ( S D = 26) , a nd th e me an numb er o f i ncorrec t trial s wa s 5 2 trial s (S D = 36 ) , leav ing a r ou n d 433 trial s per pa rtic ipan t fo r ana ly si s ( S D = 41). Quantification and statistical analyses Classification of EEG In bo t h expe rime nts , mac hine -le a rning c las s i fi cati on o f mul ticha nnel EE G d ata r eli ed on LDA a nd cros s- fold vali dati on. A s no ted a bo ve, EE G clas si fica tio n reli ed on te mpora l s earch l ights d efin ed ac r o ss a w indow of ~85 m s (17 s ampl e s i n Exp er i ment 1; 21 sa mple s in Ex perime nt 2). T r i a ls we re not av erag ed pr i or to cla s s i fic ation . For c l assi fic ati on of b oth EE G and M RI, e ach d ata f ea ture in t he training s e t (ie . voltag e a t one EEG c h ann el or a ctiva ti on o f one v oxel ) wa s z-sc o r e d acros s obs ervati on s. The mean a nd s t a nd ard de v iation v alue s deriv e d fro m normali sa tio n of th e t rai ning s et we r e r e t a in ed a nd u s e d in no rmal i s a t i o n of th e te s t set . In Experiment 1 , we ad opt ed a valid ati on scheme in c la s s i fic atio n of EE G t h at invol v ed pa r ti tionin g of trial s int o 20 chunk s , w ith 3 a nd 17 c hun ks ex hau st i vely alloc a ted in to t raini ng set s a nd t e s t se t s over 1140 itera tion s. I n Ex periment 2, trial s w ere pa rti tioned i n t o 10 ch unks , with 2 an d 8 chunk s ex haus t i vely alloc at ed into te st and t r ain i ng set s o ve r 45 iter atio ns . In bo th expe r i m ent s , th e cl assi fic atio n re sul ts re fle ct me an acc ura cy over itera tion s and pa rticipa n ts (Fi gur es 2c a nd 2 f) . Stati s tica l s i gnific anc e r e f l ect s a naly s i s o f be t w een -pa rtic ipa nt va r i anc e us i ng cl us t er- ba sed co r rec t i on over tim e with c luster d e fi nin g thre shold a t p < 0.05 and c lu ster signi f i canc e at p < 0.05 (Mari s & Oo s t e nveld, 2007 ). Cond iti onal di f f e renc e s in c la ss i fic a tion l a tencie s a r e directly interp re table , but th e ab solu te v alue s sh ould be int er p ret ed with c are a s the filte ring of EEG and u s e of t emporal s e a r c hligh ts mea nt t h a t the cla ssi fier h ad acc e ss to EE G da ta r ecorde d up to ~50 ms later than the nomina l la tenc y. Coupling variance in N2pc to variance in EEG-derived category information As d esc ribed i n th e re s u l ts secti on, we c o upled N 2pc ampl itu de t o EEG-d eriv ed t ar get c atego ry informa tion u s ing an i tera tiv e re sa mplin g approac h. Thi s ga r ne r ed 1140 v alue pai rs con taini ng mea n N2pc amplitude a nd mea n cla ssi fica t i on ac curac y for e ac h partic ip ant . The r e l atio nship b etw ee n the se va r i a ble s wa s an aly sed u sing mixed linear mo delli ng as implem e nte d in the fi t l me .m func t i o n for Ma tlab ( sta ti s tic s t o ol box R20 22a) . In mixed linear mod elling , r estric ted maxim um like lihood w as use d for va r i a nce e stima t i o n, Akai ke in fo rmation cri teri on (A IC ) wa s us ed f or mod e l co mpari s on , an d ANOVA d eriv atio ns employ ed Sa tt erthw ai t e app roxim ation s o f degr ee s o f fr eed o m. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 26 Classification of MRI As in an a lysi s o f EEG, mac hine -l e arning c l as s i fic ati on of MRI r e s u lt s labe lled th e ta rge t or di strac tor ca tegory in indivi dual t rial s u sing L DA an d c r o ss -fo ld va lidati on. S e archl ight sph er es wer e co nstrain ed when b ound by t h e edge o f brain spac e and the mea n sph er e v olume w as 111.8 voxe l s . Cla s sifica t io n wa s c omple ted u sing t w o va lidation s c h eme s. In the fir st, 4- fold c r o s s-v alida t i o n was empl oyed to g en era te t he re sul t s illu st r a ted in Fig ure 5 . S ub s e qu ently, 1 140- fold c r os s - v alida t i on wa s u sed to i den tify c ovaria nc e i n MRI cl a ssi fic ati on and N 2pc . Mean M RI cla ssific a t i on r esul ts f or 4- fold a nd 1140 - f ol d cla s si fica t i on wer e ver y simila r . Stati s tica l anal y s i s of cla s si ficati on r e sul t s employ ed t- te st s of se archli ght cla ssi fica t i on ac cur ac y ag ains t a null o f cha nce pe r fo rmanc e . Re sult s we re c lu ste r c orre cte d for mul tipl e comp arison s w it h a cl us t er de fin ing th re shold o f p < 0.001 a n d cl us t er signific anc e at p < 0 .05 (Ma ri s & O os tenve ld , 2007 ). Sta ti st i cal con tra st s of targe t v ers us di strac to r cla ssi fica tion acc uracy re lie d on s im ilar parame t e rs, bu t w ith t-t e st s c on tra sting r e s ult s from th e two a naly se s . Coupling variance in N2pc to variance in MRI-derived semantic category information Identi fic a tion o f th e r e la tion ship b etw e e n N2pc amplitude a nd MRI -de rive d ta rge t ca t e go ry i n f o r m at i o n f o ll o w e d t he s am e a nal y t i c s ch e m a as in an a ly s is of EEG fr o m Ex p er im en t 1 , w it h t he adde d dimen sion o f s e a r c hlight l oca tion . Stati stic al a naly si s of re sult s from t his an al ysi s e mploye d one- tai l ed t -te st s o f correl ation valu e s a gains t a nul l hypoth e si s of z ero . Re sult s w ere cl u s t er co r rec te d f or multipl e c ompari son s with a clust er de fining th r e s hol d of p < 0 . 001 an d clus ter signific anc e at p < 0.05. In ROI a n alysi s , region s we re de fined b a se d on re sul ts fr om se archl ight a na lysi s ill ustra ted in Fi gure 6. We appli ed our 1140 -f old c las si fica tio n t e chnique to th e vox el pa t t ern e xt r a ct e d from ea ch o f the se R OI s , gen era ting 1140 v alue p air s c ontainin g mean N 2p c and mean c la s s i f i c ation acc ura cy for eac h ROI . Fi s her - tran sform ed Pe ar s on co rrela tion s we re calc ulat ed fo r e ach R O I a nd re s u lts for e ac h ROI wer e sta tis t i call y a s s e ssed u sing a pe rmutati on te s t with 50 k iter atio ns . Sta ti s tica l anal y s i s of time-c o urse r e sult s within e a ch c lus t e r rel ied on t emp or a l cl u ster co rrec t i on with a clust er -de fining thre s ho ld of p < 0 .05 a nd clu ste r signi fic a nce at p < 0.05 . Coupling variance in N2pc to variance in MRI voxel activity Identi fic a tion o f th e r e la tion ship b etw e e n N2pc amplitude a nd vox el a ctivi t y follo we d muc h the same sche m a as d e scribed immedi a te ly ab ove but a sse s s ing vox el ac tivity r a the r tha n s e a r c hlig ht cl assi fic atio n ac curac y. I n R O I ana ly s i s, r e gion s w er e de fine d ba s ed on re sul t s illu s t ra ted in Fig ure 7 . Stati s tica l anal y s i s of the s e re sul ts r e lied on t he s a me p ermut ation te st s d e s c r i bed abov e, an d .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 27 sta t i s t ic al ana ly si s of time-c our se r e s u lts relie d on c lu ster c orrec ti on wi t h t h e sa me parame t e r s desc ribed ab ove .

Acun z o, D . J ., Low, D . M ., & F airhal l , S. L . (2022 ). De ep neur al netwo rks r evea l t opi c-l evel repr ese nt ation s o f s en ten ce s in me di al p ref ront al c orte x, la t e r a l an teri or t empora l l obe, precu neu s , and a ngular g yru s . Neur oIma g e, 251 , 11900 5. Amiez, C ., & Pe trid es , M. (200 9). Anat omic al orga niz a t io n o f the ey e fie ld s in t he h uman a nd non - huma n primat e fr ontal c o r tex . Progr e s s i n neurobiol o gy, 8 9(2), 220 -230 . Bald au f, D ., & De s imo n e, R. (201 4) . Neu ra l mec hani sms o f objec t-ba sed at t e n t i on . Sc ienc e, 344(6 182), 424 -4 27. Barbar o, L ., P ee l en, M . V. , & Hic key , C . (2 017). V alenc e , not u tili t y , und erlie s r ew a r d - d riv en prioriti z a ti on in hum an vi s ion . Jo ur na l o f Neuro s c i enc e, 37(43 ) , 10438 -10 450. Barr, D. J ., L evy, R., Sc he epe rs, C ., & Ti ly, H. J. (201 3). R andom e ffe ct s s tr uctur e fo r con f i r ma t o r y hyp ot he s i s t es t i ng: Ke e p it max imal. Jou r nal of memory and la nguage , 68(3 ), 2 55 - 278. Bar salou , L . W. (2016 ) . O n s tayi ng groun de d and a voidin g quix otic de a d end s. P syc honomic bulletin & r e view , 23, 1122 -1142. Binde r , J . R ., D esa i, R. H ., G rave s , W . W. , & Con a nt , L. L . (2009 ) . W he re i s t he se mantic s y st em? A critic al rev iew a nd me ta -an aly si s o f 120 f unc tional ne uroimagi ng st u di e s . C ere bral c or tex , 19(12 ) , 2767 -279 6. Bisl ey, J. W ., & G old b erg, M . E. (2010) . At tenti on, int enti on , and pr iori ty in the par ie t a l lob e. An nual revie w of ne ur o sci enc e, 33, 1-21 . Bracc i, S ., & de B eec k, H. O . (201 6 ). Di sso ci ation s and a s socia tio n s be t w een sh ape and c atego ry repr ese nt ation s in the two vi sual pa t hw a ys. Jo urnal o f Neu r o sci e nce, 36(2 ), 432 - 4 44. Braina rd, D. H . ( 1 997 ). The p s y chophy s i cs to olbox. Spa tial visio n, 1 0(4 ), 433-436 . Bris son, B . , Robit aill e, N., & Jolic œu r, P . (2 007). S t i mulu s in ten s i t y a ff ect s th e la te n cy b ut not the ampl itude o f th e N2pc . Neuro r e p or t , 18 (1 5), 1627 - 1 630. Broadb en t , D . E. (1958) . Pe rcepti on and co mmunica t i on. New Y or k , Pe rga mon P r ess . Brodeu r, M. B ., D i onne - Do s ti e, E., M ontr euil, T ., & Lepa ge, M . (20 10) . The Bank o f S tandard ized Stimul i (B O S S), a new s et o f 480 norma ti ve photo s o f o bject s to b e us ed a s v is u a l st i muli in co gnitive r e s e a r c h. PloS one , 5 (5) , e1077 3 . Buzsá ki, G. (201 9) . The brai n fr o m in side out. Ox ford Unive rsity Pr e s s . Cabe za, R. , Cia rame lli, E., & Mo s c ovi tch, M . (2012) . Cog ni tive cont ributio ns of t he ve nt ral pari et al co r tex : an i n t e gra tive th eor etic al a cco un t . Trend s in c ognitive s c ie nc e s , 16(6 ), 338 -3 52. Chen, L. , L ambon Ralph, M . A . , & Ro g er s , T . T. (20 17). A uni fie d model o f human s emantic kno wledg e and it s di sord er s . Na t u re h uman be ha vio ur, 1(3 ), 0039. Corb et ta, M ., & Shu lma n, G . L . (2002) . Co ntr ol o f goal -di r e c ted an d stimul u s -d rive n att ention in the brain. Natu re r ev iews ne uro scie nc e, 3 (3 ), 201-215 . Corb et ta, M ., P a tel , G. , & Shulm an , G. L. (2008). T he reor ient ing sy stem o f th e h u man brain: f rom env ironmen t to th eory o f mi nd. N euron , 58(3), 306 -324. Corb et ta, M ., P a tel , G. , & Shulm an , G. L. (2008). T he reor ient ing sy stem o f th e h u man brain: f rom env ironmen t to th eory o f mi nd. N euron , 58(3), 306 -324. De simone , R., & Dunc an, J . (19 95) . Ne ura l Mec hani sms o f Sel e ctive V i s ua l A tt e n t io n. Annual R eview of N euro s c ienc e, 18, 193–22 2. Ding, L. (202 3) . Co ntrib ution s o f th e Ba sa l Gang lia to Vi sual P erce ptua l Deci sion s . Annual Revie w o f Vi sion S cienc e , 9, 385-407. Ege t h , H. E., & Yan t i s , S . (1997 ). V i sual at tenti on: Cont rol, r epre s ent ation , an d tim e c ourse. Annu al revie w of p syc hology , 48(1) , 2 69-297. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 28 Fa irhall, S . L., & Ca ra mazz a, A . (2013 ). B r ain r e gion s th at r epre se nt amod al conc e ptual k nowled ge . Journa l of Neuro sci e nce, 33(25 ), 105 52 -1 0558. Freed man , D . J. , Rie s en h uber , M., Pog gio , T. , & Mi ll er, E. K . (2001 ). Ca te gorica l r ep r e s en ta t i o n of vi s ual s timul i in th e pr im a te pr ef r o nta l cortex . Sc ie nce , 291(55 02 ), 312-31 6 . Ga squoin e, P . G. (2014) . Con tribu tion s o f the i n s ula t o cog nition and emotion . N e urop sycho logy revie w, 24 , 77-87. Got t l ieb, J., Oudey e r , P . Y., Lope s, M ., & Baran e s, A. ( 2 013). Inf orma tion -s eeki ng, c ur i os i ty , and att entio n: c omputa tion al and ne u r a l me c hanism s . Tr end s in co gnitive s c ienc e s , 1 7(11), 585 - 593. Haxby , J. V. , Gobbi n i, M. I., Fu rey, M. L., I shai , A., Sc ho uten , J. L . , & Piet rini, P . (20 01). Di s t ribu ted and ove r l apping r e p r e s enta tion s o f fac e s and obje cts in ve n t ral tempor al c or t ex . Sc ien c e, 293(5 539), 242 5-2430 . Hayde n, B. Y. , & Niv , Y. (202 1). The ca s e ag ains t econo mic v alue s in th e or b i t o fro ntal cort ex (or any where e l s e in the br ain ). Beh avi or a l N euro s c ienc e, 135(2 ), 192. Hick ey, C ., & Pe elen , M. V . (2015 ). Neura l mecha nism s o f i ncentiv e sali enc e in na t ur a li st i c huma n vi s ion. N e uron, 85(3 ), 512 - 5 18 . Hick ey, C ., Di Lo llo , V ., & Mc Dona l d, J . J. ( 2009 ) . Elec trophy sio logic al ind ice s o f ta r get a nd di strac to r proce ssing i n visual se arc h. Journ al of c o gnitiv e neuro scie nce, 21(4 ), 760 -77 5. Hick ey, C ., G r ig nolio, D., Mu na s i nghe, V. , & Ac unz o, D . (unde r review ) . U s ing N2pc v ariabili t y to probe func tion ality : Line ar mixe d modell i ng of trial EEG and be havio u r . Hick ey, C ., Mc Donal d , J. J., & Theeuw e s , J. (2006 ). Elec trophy siol ogica l ev ide nce o f the c ap ture of vi s ual at ten tion . Jou rnal o f cogni t i ve ne u r osci ence, 18 (4), 604 - 6 13. Hick ey, C ., P olli cino, D., B erta z zoli , G. , & Barbar o, L . (2019 ). U l tra fa st o bjec t de tec tion i n nat urali st ic vi s ion r e li e s on ult raf a st di s t rac tor suppr es s io n. Jo ur n al o f C og nitive N euro scie nc e, 3 1(10) , 1563 -1572. Hüer, J ., S axena , P ., & Tr eue, S . (202 4) . P athwa y- s e l ectiv e op togen e t i c s reve al s th e fun cti onal ana t o my of top –dow n at ten tional modul ation in th e ma caque vi sual cor tex. P roc e eding s o f the Na tion al Acade my o f S cienc e s, 121(3 ), e230451 1121. Humphreys , G. F ., Ralp h, M . A . L., & S imo ns, J. S. (20 21). A unify ing a ccou nt of ang ula r gyrus co nt ribu tion s to ep i s odi c an d s e man tic c ognition . Tr end s in Neu ro s c ienc e s, 44(6) , 452 - 4 63. Jeong , S . K., & X u, Y. (20 16 ). B ehav ior ally r e leva n t ab s tr ac t objec t i d enti ty repr e se ntati on in the huma n parie tal cor tex. Jo urnal o f Ne ur o s ci ence , 36(5 ), 16 07-1619 . Klei n-Flügg e, M . C ., Bongio a nni, A ., & Ru shwor th, M. F . (2022 ). Me di al and orb it a l fron tal c or t ex in dec isi on-maki ng and f l exibl e b eha vior. N euron, 110 (17 ), 2743 -277 0. Kriege s k or t e, N., G o eb el , R., & Band ettin i , P. (200 6 ). I nfo rmati on-b a sed func tion al brai n mapping . Proce eding s of the N ation a l Aca demy o f S cie nce s, 103 ( 1 0 ), 3863-38 68. Le vy, D . J . , & Gli mche r, P . W . ( 2 012) . The root o f all v alue : a neur al common curre ncy for choic e. Curr ent opi nion in neu robio logy, 22(6 ), 1 027 - 1 038. Lu ck, S. J., & F ord, M . A . (1998). On th e r ole of s el ectiv e a tt entio n in visual pe rc ep tion. Proc e eding s of th e Na tiona l Ac ad emy of S cie nc e s, 95( 3), 825-830. Lu ck, S. J., & Hilly a rd, S. A . ( 1 994 ). El ect ro physiolog ical cor rela te s o f f eatu re a n alys is during v isua l se arch. P s y chophy s iol ogy, 3 1(3), 291 -30 8 . Lu ck, S. J., Gi relli , M ., Mc Dermo tt, M. T., & F or d, M . A . (1997) . Br id ging th e gap be t w een mo nkey neurophy s iol ogy a nd huma n p ercep tion : An ambig uity re solu tion th eory o f visual sel ec tive att entio n. C o gnitive p s y cholog y, 33 (1), 6 4-87. Lu r ia , R., B a laban , H ., Awh, E. , & Vog el, E . K. (2016 ). The c ontra lat eral d e lay a ctivi ty as a ne ura l mea s u re o f v is u al wo r k ing me mor y . Neu ros c i enc e & Biobeh av ioral R ev iews, 62 , 1 00-108 . Mahon, B . Z., & Caram a z za, A. (2 009 ). C o nc ept s a nd ca te gori es : A cog nitive n euro ps y cholog ica l per spec tiv e. Annu al r e view o f psycho log y, 60, 27-51 . Mari s, E., & Oo stenv eld , R. (2007 ). N onp arame tric sta ti stical t e s ting o f EEG- and MEG-d ata . J o u r na l o f neuro scie nc e me t h o ds, 1 64(1) , 177-1 90. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 29 Martin , A. (201 6). GRAP ES— G roundi ng repre s enta tion s in ac ti on, pe rce pti on, and emotion sys t e m s : How ob ject prope rti es and c at egorie s a r e r ep re sen ted i n th e human bra in. P syc h onomic bull etin & r ev iew , 23, 97 9 -990. Maun sell, J . H . (2015). Neu ronal mecha ni s m s of v i s ua l a tte ntion . Annu a l revie w of v is i on s c i enc e, 1, 373-39 1. Menon, V., & U ddin , L. Q . (201 0 ). Sal i enc y , switchi ng, at ten tion a nd co n t rol : a ne t work model of insula fun cti on. B rain s truc tur e an d fu nct ion, 214, 65 5-667. Mille r , E. K. , & C ohe n , J. D. (2001 ). A n int egrative t he o r y of pre fr onta l c or t ex func tion. A nnual r ev iew of neu ro scie nce , 24(1) , 167-2 02. Niv, Y . (2019). Le arning ta s k - s t ate r epr e s enta tion s. Natu r e ne uro s c ienc e, 22(10 ) , 1544-155 3. Pado a -Sch iop pa, C . , & C on en, K. E. (2017). Orbito fro nt a l co rtex : a n eu r a l circ ui t fo r econom ic dec isi on s . Neuron , 96(4) , 736-7 54. Pel le tie r, G ., & F ello w s , L. K. (2021) . Vi e wing or b it o fron tal cor tex c ontrib ution s t o decisi on-ma king through th e l en s o f objec t rec ogni t io n . B ehav ioral N euro s c i enc e, 135 (2 ), 182. Pome r l e au, V. J., F o rt i e r-Gau thie r, U. , C o r rive au, I ., D e ll'Acq ua, R ., & Jolic œur , P. ( 2014). Col our - speci fi c dif fe renc e s in a tt enti onal depl oy ment fo r equilu min ant pop -ou t c olour s: Eviden ce from la ter ali sed po ten tial s. In te rnat ional Jour n a l of Psyc hophy siol og y, 91(3), 194 - 205. Po sner , M. I ., Snyde r , C . R., & David son , B. J . (1980) . At ten tion an d the de tec t i on of signal s. J ournal of ex perim ental p s y cholog y: G ene ral, 10 9(2), 16 0. Princ e, J. S ., Ch are s t , I. , Kurz a w ski, J. W., P yle s, J . A., T a rr , M. J., & Kay, K . N . (2022 ). Impr ov ing the ac curac y of singl e -t ri al fMR I re s p on se e st imate s u sing GL Msingl e. Eli f e , 11, e 7759 9. Rash al, E., S e nou ssi, M ., San tandr ea , E. , Be n- H am ed, S ., Maca lu so, E., Chel azzi, L ., & Boehl er, C . N. (2022 ) . An EEG s t ud y of the c ombine d ef fect s o f to p-d own a nd bot tom- up at ten ti onal sel ec t i on und er va r y ing ta sk di fficul ty. P sy chophy s i ology, 59(6 ), e1 4002. Rey nold s, J . H., & H ee ge r, D . J . (2009 ). The no r ma liz a t i on mod el of a tt entio n. N eur on, 61(2) , 168- 185. Ric e, G. E., W a tson, D. M. , Har tley , T., & Andrew s, T. J. (2014 ). L ow-l ev el im age pr ope r t ie s of v i s ua l objec t s p r ed ic t pat ter ns o f ne ural r e spon se ac r o s s categ o r y - s el ective r egion s of th e v entr a l vi s ual pa thwa y. Jou rnal o f N eu ro scie nce , 34(26), 883 7-8844 . Rigo t ti, M., Ba rak, O., Wa rden , M. R. , Wa ng, X. J., D aw , N. D., Mill er, E. K., & Fu si, S . (2013) . T he importanc e of mix ed selec tivity i n comp lex c ognitiv e ta sks . N a t u r e , 497(7451 ), 585-590 . Ruec kl, J . G ., Paz-Alo n so, P . M., M olf e se, P. J., Kuo, W . J. , Bi ck, A . , F ro s t, S . J. , .. . & F r os t, R . (201 5) . Univer sal brai n s i gna tur e of pr ofic i ent r e ading : Evide nce f r om fou r c ontra s ting la ngu ages . Proce eding s of the N ation a l Aca demy o f S cie nce s, 112 ( 5 0 ), 15510 - 1 5515. Sa wak i, R., & Lu ck, S. J. (201 0 ). Captu re v e rsu s s u ppre s sion o f a tt ention b y salie nt singl e t o n s : Ele ctrophy siologi cal evid e nce f or an au to matic att end -to - m e s i gnal . A t ten tion, Per ception, & Psyc hophy sic s, 72(6 ), 1455-147 0 . Sc huck , N. W., Cai, M . B. , Wil son, R . C . , & Niv , Y. (201 6). Human o rbit of ront al c or t ex repre s e nts a co gnitive map of st ate sp ace . Ne ur o n, 91 (6), 14 02 -1412. Se eley, W. W . , Meno n, V ., S cha tzberg, A. F ., Kel ler , J., Glove r, G. H ., K enna , H., . . . & Gr eic iu s , M. D. (2007 ) . D i ssocia ble in trin sic conn ec tivity n etwork s f or s al ienc e proce s s i ng and exe c utive co nt rol . J o u r na l o f neur osci e nce, 27 (9), 2 349-2 356. Se idl, K. N., P eelen , M. V., & Ka st n e r , S. ( 2012 ) . N e ur a l evi d ence f or di s tr ac t e r s up pre s s ion during vi s ual s e a r c h in r e al -world sce n e s . Jou r n a l of N euro s c ie nc e, 32(34 ), 11812-11 819 . Se stie ri, C ., Shulma n, G. L. , & C o r be tt a, M. (2017) . The cont ributi on o f t he huma n p osteri or pari eta l co r tex to e pis odic memory. N atur e Rev ie ws Neu ro s c ienc e , 18(3 ), 183-192. Stoja no s k i, B. , & Cu sack , R. (201 4 ). Time t o wav e good-by e to pha s e scr amb ling: C rea t i ng co nt roll ed scrambl e d imag e s u sing dif feomo rphic tra nsform ation s . Jou rnal o f v ision , 14(12 ), 6-6. Uddin, L . Q., Nomi, J . S. , Hé ber t -S erop ian , B. , Gh az i r i, J . , & Bo uch er, O . (2017 ) . St ru cture and fun cti on of th e hu man in sula . J o urna l of Clin ical Ne ur op hy s i ology , 34(4 ), 30 0-306 . .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint 30 Ungerleid er , S . K. A . L. G . (20 00 ). Mech a n ism s of v i sual a tte ntion i n the huma n cor tex. Annua l rev iew of neu ro scie nce , 23(1) , 315-3 41. Verl ege r, R., ve l G r a jew ska, B. Ż., & Ja śko wski , P. (201 2) . Time-c ou rse o f hemi sphe ric pr e fe ren ce f or proce ssing c ontra lat eral r ele van t sh ape s : P1pc , N1pc, N 2p c, N3pc. Advanc e s in c o gnitive psych ology, 8(1 ), 19. Voge l, E. K. , & Ma chi zawa , M. G. (2004). N eur al activ ity predic t s in dividua l dif fe re nc es in vi sual work ing mem or y ca pacity . Na tur e , 428(6 984 ) , 748-751 . Wagne r , A . D. , Shan n on, B . J. , Ka hn, I. , & Buc kner, R. L. (200 5) . Pari et al lobe c on tr i bu t i on s to e pi s od ic memory r e t r iev al. T re nd s i n cog nitive sci ence s, 9(9 ), 445-453 . Wat s on , D. M ., Ha rt l ey, T . , & A n d r e w s , T. J. (201 4) . P att ern s o f r e sp on se to v i s ua l s c en e s are l inke d to the l ow -leve l p rope rtie s o f th e imag e. Neuro Image , 99, 4 02 -410. Weav er, M. D ., va n Zoe st , W., & H ic key , C. (20 17 ). A tempor al de p endenc y a ccou n t of a tte ntio nal inhi bition in o culomo tor c on t rol . N e u r oI ma ge, 1 47, 880 - 8 94. Woodman, G . F. , & Luc k, S . J . (2003 ). Seri al deploy ment o f a tte ntion du ring visual se arc h. Journ al of Ex perimental P s y chol og y: Human P e r c ep tion a nd P er forma nce , 29(1 ), 1 21. Wu, C . Y ., Ho, M . H. R. , & C hen, S . H. A . ( 201 2). A met a- ana ly si s of fMR I studie s o n Chine se orthogr aphic , phon ol ogica l, and s em an ti c proc essing . N e u r oi mage, 63 (1), 381 -3 9 1. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 29, 2024. ; https://doi.org/10.1101/2024.05.29.596408doi: bioRxiv preprint