task Cognitive, Affective and Behavioral Neuroscience, Sep !!" #y Ansari, $ahereh %, &erakshan, Na'anin, (ichards, Anne According to the attentional control theory of anxiety (Eysenck, Derakshan, Santos, & Calvo, 2007), anxiety i!airs !erforance on cognitive tasks that involve the shifting f"nction of #orking eory$ %his hy!othesis #as tested "sing a ixed antisaccade !aradig, in #hich !artici!ants !erfored single&task and ixed&task versions of the !aradig$ %he single task involved the co!letion of se!arate 'locks of anti& and !rosaccade trials, #hereas in the ixed task, !artici!ants co!leted anti& and !rosaccade trials in a rando order #ithin 'locks$ Analysis of s#itch costs sho#ed that high&anxio"s individ"als did not exhi'it the coonly re!orted !aradoxical i!roveent in saccade latency, #hereas lo#&anxio"s individ"als did$ %he findings are disc"ssed #ithin the frae#ork of attentional control theory$ %here is 'ehavioral evidence to sho# that anxiety is associated #ith an attentional 'ias for threat&related aterial (see (ar&)ai, *ay, +ergain, (akerans,ranen'"rg, & van -.endoorn, 2007, for a revie#)$ /ecent ne"roiaging #ork has also sho#n that anxiety selectively facilitates early !rocessing of threat and enhances distracti'ility to task&irrelevant sti"li$ According to (isho! (20070 see also (isho!, D"ncan, (rett, & *a#rence, 2001), anxiety is associated #ith enhanced aygdala activation and red"ced recr"itent of !refrontal cortical areas (es!ecially the dorsal lateral !refrontal cortex 2D*+3C4 and the ventral lateral !refrontal cortex 25*+3C4) that are heavily involved in to!&do#n reg"lation of attention, es!ecially #hen attentional foc"s is re6"ired for efficient task !erforance$ (oth 'ehavioral and ne"roiaging #ork has sho#n that anxiety is associated #ith adverse effects on cognitive !erforance, es!ecially on tasks that re6"ire attentional foc"s$ 7n an atte!t to ex!lain the role of attentional control in anxiety and cognitive !erforance, the attentional control theory of anxiety #as !"t for#ard 'y Eysenck, Derakshan, Santos, and Calvo (2007)$ %he attentional control theory (Eysenck et al$, 2007) is a a8or develo!ent of !rocessing efficiency theory (Eysenck & Calvo, 9::2)$ (ased on (addeley;s (9:<=0 see also Derry'erry & /eed, 2002) #orking eory odel, the theory clais that anxiety disr"!ts the 'alance 'et#een #hat Cor'etta and Sh"lan (2002) disting"ished as the sti"l"s&driven (involved in 'otto&"! control, infl"enced 'y salient environental sti"li) and the goaldirected (involved in to!&do#n control, infl"enced 'y the c"rrent goal) systes$ %hese t#o systes are generally tho"ght to interact in their f"nctioning (+ashler, >ohnston, & /"thr"ff, 2009), '"t anxiety is 'elieved to increase the infl"ence of the sti"l"s&driven syste over the goaldirected !rocesses, red"cing attentional control$ +redictions of the attentional control theory are 'ased on a f"ndaental distinction 'y the !rocessing efficiency theory (Eysenck & Calvo, 9::2) 'et#een !erforance effectiveness and !erforance efficiency$ Effectiveness refers to an individ"al;s co!etence in doing a task (eas"red 'y res!onse acc"racy), and efficiency refers to the ao"nt of !rocessing reso"rces invested in doing the task (eas"red 'y res!onse latency)$ %he theory !redicts that anxiety has a greater i!act on !erforance efficiency of tasks re6"iring the inhi'ition (?one;s a'ility to deli'erately inhi'it doinant, a"toatic, or !re!otent res!onses #hen necessary?0 @iyake et al$, 2000, !$ A7) andBor the shifting (?shifting 'ack and forth 'et#een "lti!le tasks?0 @iyake et al$, 2000, !$ AA) f"nctions of the central exec"tive$ 7n inhi'ition, attentional control !revents attentional reso"rces fro 'eing allocated to task&irrelevant sti"li, and in shifting, attentional control is "sed in a !ositive #ay to allocate attentional reso"rces to exec"te the task relevant to the c"rrent goal$ A general ass"!tion of the attentional control theory (Eysenck et al$, 2007) is that attentional control involves the "se of the !rocessing reso"rces of the central exec"tive$ )o#ever, an alternative !ro!osal is that there are rather s!ecific reso"rce !ools and that those involved in attentional control differ fro those involved in other !rocessing f"nctions of the central exec"tive (Cickens, 9:<1)$ %he ass"!tion #ithin attentional control theory is ore !arsionio"s than the one 'ased on a "lti!lereso"rce a!!roach$ )o#ever, it is acce!ted that f"rther research #ill 'e re6"ired to el"cidate the "nderlying echaniss involved$ Derakshan, Ansari, )ansard, Shoker, and Eysenck (in !ress) exained the relationshi! 'et#een anxiety and inhi'itory !rocesses, "sing the antisaccade !aradig ()allet, 9:7<)$ 7n this task, to!&do#n attentional control is exercised to s"!!ress a reflexive saccade to#ard an a'r"!t !eri!heral sti"l"s (i$e$, inhi'it) and to generate a volitional saccade to its irror !osition (antisaccade)$ +erforance on this task is co!ared #ith that on the !rosaccade task, #hich eliinates the conflict 'et#een reflexive and volitional !rocesses 'y re6"iring !artici!ants to look at the a'r"!t c"e #hen it a!!ears$ As #ell as aking ore directional errors, individ"als are generally slo#er to ake a correct saccade a#ay fro the a'r"!t sti"li on antisaccade trials, as co!ared #ith !rosaccade trials (see )"tton & Ettinger, 200=, for a revie#)$ 7n the antisaccade task, co!etition occ"rs 'et#een reflexive saccades (!re!otent res!onse) and the antisaccade (volitional res!onse) (@assen, 2001)$ %o !redict correct antisaccade !erforance, Ettinger et al$ (200<) identified s!ecific 'rain regions involved in attentional control$ 7n a ne"roiaging st"dy, they fo"nd enhanced activation in the s"!raarginal gys, D*+3C, and 5*+3C on correct antisaccade trials and inter!reted the activation in these areas as indicative of a a8or involveent in the inhi'ition of reflexive saccades$ 3"rtherore, it #as fo"nd that the D*+3C and the 5*+3C #ere also involved in volitional saccade generation Derakshan et al$ (in !ress) re!orted that high&anxio"s ()A) individ"als had longer correct antisaccade latencies than did lo#&anxio"s (*A) individ"als '"t that the t#o gro"!s did not differ on the !rosaccade task, s"ggesting that anxiety is associated #ith the inhi'itory co!onent of attentional control$ %here #as no difference 'et#een the t#o gro"!s in ters of saccadic error rate$ %hese res"lts indicated that anxiety affects efficiency and not effectiveness$ Diven the involveent of the D*+3C and the 5*+3C in correct antisaccade !erforance, it is reasona'le to ass"e that these areas #ill 'e ost affected 'y anxiety and its od"lation in attentional control, as assessed 'y the antisaccade task$ S!ecifically, these !refrontal areas have 'een arg"ed to 'e involved not only in inhi'ition, '"t also in f"nctions s"ch as the re!resentation of task set #hen t#o or ore tasks co!ete #ith one another for exec"tion (Daravan, )ester, @"r!hy, 3ass'ender, & ,elly, 200=0 )erath, ,ling'"rg, Eo"ng, A"nts, & /oland, 2009)$ Altho"gh the !resent st"dy did not "se f@/7 ethods, the involveent of these areas in antisaccade !erforance allo#s a relia'le inter!retation of saccadic eye oveents as an index of attentional control$ %he 'asic ass"!tions are in line #ith the !redictions of attentional control theory (Eysenck et al$, 2007)$ %he ai of the !resent st"dy #as to test the theoretical !rediction that anxiety affects efficient task&s#itching !erforance, #hich involves reconfig"ration of task sets (!araeters associated #ith each task) in #orking eory to allo# exec"tion of the task a!!ro!riate to the c"rrent goal (e$g$, /"'instein, @eyer, & Evans, 20090 for a revie#, see @onsell, 200F)$ %he vie# that anxiety affects task&s#itching !erforance is 'ased on t#o ain ass"!tions$ 3irst, according to Eysenck et al$ (2007), anxiety affects !erforance on tasks that involve ra!id shifting 'et#een s"ccessive task sets (aintained and "!dated in #orking eory) d"e to the adverse effect of anxiety on to!&do#n attentional control (Eysenck, +ayne, & Derakshan, 200A0 )arris & diing, 200F)$ %he second ass"!tion relates to the i!ortant role of #orking eory ca!acity in s"ccessf"l task set reconfig"ration$ @ore s!ecifically, the a'ility to s"ccessf"lly select and exec"te the goal&relevant task set (#hile inhi'iting or terinating the irrelevant) re6"ires efficient "se of #orking eory reso"rces to exercise to!&do#n attentional control ((addeley, Chincotta, & Adla, 20090 )ester & Daravan, 200A0 ,ane, (leckley, Con#ay, & Engle, 20090 @eiran, 9::=)$ Daravan et al$ (200=), "sing f@/7, fo"nd that the D*+3C had a s"!ervisory role in #orking eory f"nctions, s"ch as re!resentation of task set, sti"l"s a!!raisal, and res!onse selection$ (ehavioral data ()ester & Daravan, 200A, Ex!erient 9) sho#ed that increasing the n"'er of ites to 'e held in #orking eory decreased the s!eed at #hich !artici!ants s#itched attention fro a !riary #orking eory task to a secondary decision&aking task$ %his s"ggests that diinished #orking eory ca!acity decreases the efficiency #ith #hich attentional control can 'e exerted %o test this !rediction systeatically, #e ado!ted the ixed antisaccade task to exaine the effect of anxiety on task&s#itching !erforance$ %his !aradig re6"ires !artici!ants to randoly s#itch 'et#een anti& and !rosaccade tasks in the ixed&task 'lock, #hereas in the singletask 'lock, they !erfored either the antisaccade or the !rosaccade task se!arately$ S#itching !erforance is assessed 'y co!aring !erforance on trials in #hich !artici!ants !erfor one task re!eatedly (in the single&task 'lock, or re!eat trials in the ixed&task 'lock) #ith trials in #hich they are re6"ired to randoly alternate 'et#een t#o tasks (ixed&task 'lock at a generalBglo'al level or s#itch trials in this 'lock at a s!ecificBlocal level0 see, e$g$, ,ray, 200=)$ %here is a !aradoxical i!roveent in antisaccade !erforance #hen !artici!ants are re6"ired to s#itch 'et#een anti& and !rosaccade trials, as co!ared #ith re!eat trials$ 3or exa!le, Cherkasova, @anoach, 7ntriligator, and (arton (2002) exained resid"al s#itch cost in a randoi.ed ixed antisaccade task (co!aring latencies and error rates 'et#een the s#itch and the re!eat trials, in the ixed&task 'lock)$ %hey o'served red"ced antisaccade latencies and greater error rates in the s#itch trials '"t an effect that #as not attri'"ta'le to a s!eed&acc"racy tradeoff (see also @anoach et al$, 2002)$ )odgson, Dolding, @olyva, /osenthal, and ,ennard (2001, Ex!erient 9), "sing a ore !redicta'le !resentation se6"ence (a s#itch after eight consec"tive trials of one saccade ty!e), co!ared antisaccade !erforance in the single&task and ixed&task 'locks$ %hey re!orted a significant red"ction in antisaccade latencies in the latter, #hich they s"ggested ay have 'een d"e to allocation of ore attentional reso"rces to the task in this 'lock, ?leading to a 'enefit in !erforance of antisaccade task? (!$ F20)$ %his !aradoxical i!roveent can 'e inter!reted #ithin a goal&driven controlled !rocessing a!!roach (see Gie"#enh"is, (roerse, Gielen, & >ong, 2001), so that ?!erforance varia'ility in tasks assessing exec"tive f"nctions ay arise fro fail"res to f"lly or consistently foc"s attention on task re6"ireents $ $ $ altho"gh task re6"ireents ay 'e "nderstood and ree'ered, they are not t"rned into active goals or ade6"ately aintained as s"ch? (!$ 9::), also kno#n as goal neglect (De long, (erendsen, & Cools, 9:::)$ )ence, the !aradoxical s#itch 'enefit in anti saccade latency co"ld 'e d"e to the allocation of ore attentional reso"rces to the task re6"ireents #hen the task goal is externally !resented and, hence, "!dated on every trial (in the ixed&task 'lock), as co!ared #ith #hen the task goal is !resented at the 'eginning of a 'lock of consec"tive trials (in the single&task 'lock) and "st 'e aintained thro"gho"t and activated internally$ )ere, #e exained ho# anxiety affected the o'served i!roveent in s#itching efficiency on the antisaccade task$ Saccade latency and acc"racy in the ixed&task 'lock #ere co!ared #ith !erforance in the single&task 'lock$ Saccade !erforance #as also co!ared 'et#een the s#itch and the re!eat trials #ithin the ixed&task 'lock$ Hn the 'asis of the !redictions of the attentional control theory (Eysenck et al$, 2007), #e ex!ected that )A individ"als, as co!ared #ith *A individ"als, #o"ld not sho# the i!roved antisaccade !erforance, as eas"red 'y s#itch cost$ 7n other #ords, #hen shifting #as involved, anxiety #o"ld interfere #ith the efficient allocation of attentional reso"rces to i!leent the c"rrent goal relevant to the task, as eas"red 'y correct saccade latency$ )o#ever, #e !redicted that anxiety #o"ld not i!air !erforance effectiveness, as eas"red here 'y saccadic error rate$ @E%)HD +artici!ants 3ifty&nine !artici!ants (ean age I F9$:, in I 9<$00 ax I 1=, SD I 7$A) #ere recr"ited 'y eans of advertiseent and "niversity !artici!ant !anels$ All had corrected&to&noral vision and #ere allo#ed to #ear their glasses or contact lenses$ Each individ"al #as tested in a se!arate la'oratory session lasting a!!roxiately 10 in and #ere rei'"rsed ( A) for their contri'"tion$ C"e and %arget Sti"li %he sti"l"s serving as c"e #as an oval&sha!ed o'8ect (F$F = ), and the target sti"l"s #as an arro# !ointing either "! or do#n (0$= 2$F )$ %hese sti"li #ere created "sing the @icrosoft +aint gra!hics a!!lication$ Eyetracking S!ecification Eye oveents #ere eas"red "no'tr"sively via a reote caera o"nted 'elo# the co!"ter onitor and "sing the *C %echnologies Eyega.e syste (*C %echnology, 7nc$, 200F) to track the !artici!ant;s eye oveents$ %his is an infrared&'ased eyetracking soft#are that generates ra# ga.e location data at a sa!ling rate of =0 )., allo#ing eye oveents to 'e recorded every 9=$=7 sec$ 7t "ses the !"!il¢er corneal reflection ethod to estiate ga.e !oints (i$e$, the intersection of the o!tic axis #ith the screen)$ %he eyetracker is cali'rated "sing nine fixation !oints, #ith #hich the soft#are indicates #hether or not valid ga.e !oints can 'e calc"lated$ %he Eyega.e syste accoodates several so"rces of error, s"ch as head range variation and !"!il diaeter variation$ Data are collected fro the eye for #hich the ost ra!id and acc"rate cali'ration is o'tained$ %he !resentation of the sti"li #as controlled 'y the D@DJ !rogra (3orster & 3orster, 200F), #hich also recorded res!onses fro the '"tton 'ox (+7H&92 interface) at 9000 )., ens"ring illisecond tiing acc"racy$ %he sti"li #ere !resented in 21&'it color on a 9,021 7=< *CD onitor (5ie#Sonic 700'0 cell res!onse tie, FA sec)$ %he eyetracking syste and D@DJ #ere a"toatically synchroni.ed at the 'eginning of each trial$ Ex!eriental %ask %he !artici!ants #ere re6"ired to co!lete antisaccade and !rosaccade tasks !resented in t#o 'locksK single task and ixed task$ 7n the single&task 'lock, the !artici!ants co!leted trials of the sae task, either anti& or !rosaccade, consec"tively in different 'locks$ 7n the ixed&task 'lock, anti& and !rosaccade trials #ere !resented randoly, re6"iring the !artici!ants to s#itch 'et#een the t#o tasks$ %hese tasks #ill 'e detailed 'elo#$ And& and !rosaccade tasks$ As is de!icted in 3ig"res 9A and 9(, a trial 'egan #ith the #ord ?/eady,? follo#ed 'y a central fixation sy'ol (s"'tending 0$= )$ %he !artici!ants #ere instr"cted to fixate it "ntil it disa!!eared$ 7n the re!eat condition (3ig"re 9A), this fixation sy'ol #as a #hite cross (2 2 )$ 7n the s#itch condition (3ig"re 9(), the fixation sy'ol #as either a #hite diaond (9$20 2$1 ) or a #hite circle (9$7 diaeter) and !rovided instr"ctions for the ty!e of saccade to 'e !erfored on that trial (e$g$, the diaond ight indicate a !rosaccade trial, and the circle an antisaccade trial)$ After sy'ol offset and a 200&sec ga!, an oval o'8ect a!!eared 99 to either the left or the right side of the screen$ %he !artici!ants #ere re6"ired to direct their ga.e either ?a#ay? (antisaccade) fro the oval c"e to its irror !osition on the screen (#itho"t looking at the c"e) or ?to#ard? it (!rosaccade)$ At c"e offset, an arro# either re!laced the c"e (on !rosaccade trials) or a!!eared on the side of the screen o!!osite to the c"e (on antisaccade trials)$ %he !artici!ants #ere re6"ired to identify the direction of the arro# ("! or do#n) 'y !ressing the relevant '"tton on the '"tton!ress 'ox$9 %he target arro# #as sho#n for 900 sec$ Ex!eriental +roced"re %he !artici!ants co!leted the ex!erient in t#o sessions se!arated 'y a short 'reak$ 7n half of the ex!eriental 'locks, anti& and !rosaccade trials #ere !resented randoly (ixed&task 'lock), and in the reaining half, anti& and !rosaccade trials #ere !resented consec"tively in se!arate 'locks (single&task 'lock)$ 7n the single&task 'lock, the !artici!ants #ere instr"cted as follo#sK 7n this task, yo" #ill 'e asked to either look %HCA/DS or ACAE 3/H@ an o'8ect #hich #ill a!!ear either on the left or right side of the screen$ Every trial #ill 'egin #ith the #ord ;/eady; follo#ed 'y a cross ( )$ Chenever this a!!ears look at it$ Chen asked to look to#ard the o'8ect, yo" "st look A% it 5E/E L-7C,*E #hen it a!!ears$ Chen asked to look a#ay, yo" @-S% GH% look at the o'8ect '"t look ACAE to the side of the screen o!!osite the oval o'8ect 5E/E L-7C,*E$ An arro# #ill then a!!ear on the screen, yo" sho"ld identify the direction of the arro# (!ointing -+ or DHCG) "sing the a!!ro!riate keys on the 'ox$ Eo" sho"ld do this 5E/E L-7C,*E$ So ree'er that S!eed and Acc"racy are very i!ortant$ And this #as the sae for the ixed&task 'lock, #ith the follo#ing odificationK $ $ $ %his 2the cross4 #ill 'e follo#ed 'y a sy'ol that can either 'e a D7A@HGD or a C7/C*E$ %hen an oval&sha!ed o'8ect #ill 'e flashed on either the left or right side of the screen$ Chen the sy'ol is a circle yo" "st look ACAE to the side of the screen o!!osite the oval o'8ect 5E/E L-7C,*E (yo" @-S% GH% *HH, at the oval o'8ect at all)$ Chen the sy'ol is a diaond then yo" "st look A% the oval o'8ect as soon as it a!!ears$ An arro# #ill then$ $ $ %he !artici!ants #ere told that the !"r!ose of the st"dy #as to exaine !erce!tion of o'8ects$ After co!leting the consent for, the !artici!ants rated their ood on five 900& vis"al analog"e scales (5ASs0 0, not at all, to 900, very "ch) to indicate their ood states at that !oint in tie (e$g$, (ond, Shine, & (r"ce, 9::A)$ %he diensions incl"ded ha!!y, irrita'le, sad, anxio"s, and coforta'le states$ %he !artici!ants co!leted these scales at the end of the ex!erient also$ Gext, half of the !artici!ants co!leted the eas"re of trait anxiety, the State&%rait Anxiety 7nventory (S!iel'erger, Dors"ch, *"shene, 5agg, & >aco's, 9:<F), after #hich they #ere given a distractor task involving searching for vo#els in different atrices of letters$ %his lasted "nder A in$ %he reaining !artici!ants co!leted the scale at the session;s end$ %he !artici!ants #ere given an overvie# of the eyetracking e6"i!ent, and the res!onse '"ttons #ere indicated$ %hey #ere seated directly in front of the co!"ter onitor, #ith their chin !laced on a chinrest (located =0 c fro the screen), in a dily lit c"'icle$ %#enty&fo"r !ractice trials #ere co!leted (92 s#itch, 92 re!eat), at the end of #hich the !artici!ants #ere asked to ver'ali.e the instr"ctions in order to ens"re that they had 'een f"lly "nderstood$ %he eyetracker #as cali'rated$ S!eed and acc"racy #ere e!hasi.ed$ At the end, the !artici!ants #ere f"lly de'riefed on the !"r!ose of the st"dy and #ere !aid for their contri'"tion$ %he ex!erient co!rised eight 'locks of 20 ex!eriental trials each, half of #hich #ere antisaccade and half of #hich #ere !rosaccade (F20 trials in total, half of #hich #ere s#itch and half re!eat trials)$ %he orders in #hich the anti& and !rosaccade trials (A(& (A) and the single&task and ixed&task 'locks (A((A&(AA() #ere !resented #ere co"nter'alanced in a 'et#een&s"'8ects design$ %he instr"ctional eaning of the circle and diaond fixation sy'ols "sed in the ixed&task trials #as also co"nter'alanced0 hence, on half of the trials, the diaond sy'ol instr"cted an antisaccade (and the circle a !rosaccade), #hereas on the reaining half, it instr"cted a !rosaccade (and the circle an antisaccade)$ Data +re!aration Saccades #ere defined as eye oveents #ith velocities exceeding F0 Bsec (@assen, 20010 /e"ter, > ger, (ottlender, & ,athann, 2007) and a!lit"des exceeding F that #ere ade after c"e onset and 'efore c"e offset$ Ce exained directional acc"racy (!ercentage of error), #hich de!ended on the re6"ired res!onse (antisaccade or !rosaccade), and latency of the first correct saccade$ An incorrect saccade #as defined as the first saccade after c"e onset #ith an a!lit"de MF to#ard the !osition of the c"e (on antisaccade trials) or a#ay fro the c"e (on !rosaccade trials)$ %he latency of the first correct saccade #as defined as the ela!sed tie 'et#een the onset of the c"e and the 'eginning of a saccade to#ard the correct area of interest$ Ce exained task&s#itching !erforance 'y first co!aring saccade latency and error rate in the single&task 'lock #ith those in the ixed&task 'lock (in #hich anti& and !rosaccade !resentation #as randoi.ed)$ Ce calc"lated latency s#itch cost 'y s"'tracting ean correct saccade latency in the single&task 'lock fro ean correct saccade latency in the ixed&task 'lock, a coonly "sed ethodology$ Cith this co!arison, it is diffic"lt to disentangle the !ossi'le contri'"tions of c"ing of task goal in the ixed&task 'lock$ 7n order to avoid this !ossi'le confo"nd, #e also !erfored trial& 'y&trial analysis in the ixed&task 'lock$ 7n this 'lock, trials #ere classified as either s#itch (those trials !receded 'y a trial of the different task ty!e) or re!eat (those trials !receded 'y a trial of the sae task ty!e)$ *atency s#itch cost in the ixed&task 'lock #as calc"lated 'y s"'tracting the ean latency on re!eat trials fro that on s#itch trials$ Error rates #ere s"'8ected to the sae !roced"re$ %rials #ere excl"ded fro analysis if eyetracking #as interr"!ted d"e to a lost !"!il or if no eye oveents #ere ade 'y the !artici!ant$ %rials on #hich the onset of the first saccade #as shorter than <F sec (i$e$, antici!atory) or the latency to res!ond #as greater than =00 sec #ere also excl"ded fro analysis$ %hese criteria res"lted in the loss of 1$F9N of the antisaccade and 1$F<N of the !rosaccade trials in the s#itch condition and 1$AAN of the antisaccade and 1$9=N of the !rosaccade trials in the re!eat condition$ An analysis sho#ed that data loss did not differ as a f"nction of task or trial ty!e$ (ehavioral Data %he !artici!ants #ere also eas"red on reaction tie of correct res!onse to arro# identification$ %rials #ith incorrect res!onses #ere reoved, as #ere trials #ith reaction ties less than 900 sec or greater than 2,000 sec$ Hf the data, A$2N #ere lost d"e to o"tliers and erroneo"s res!onses$ /ES-*%S +artici!ants 3o"r !eo!le had to 'e excl"ded d"e to !oor tracking$ %his res"lted in AA !artici!ants in the final data set$ @ean trait anxiety score as eas"red 'y the S!iel'erger State&%rait Anxiety L"estionnaire (S!iel'erger et al$, 9:<F) for the sa!le#as 12$12 (SD I <$:9, in I 2F, ax I =1)$ Ce !erfored tercile s!lits on the trait anxiety 6"estionnaire to !rod"ce lo# and high extree gro"! classifications$ %he *A individ"als (n I 20) incl"ded those scoring F< or lo#er, and the )A individ"als (n I 29) incl"ded those scoring 1A or higher on this scale$ @ean trait anxiety for the *A gro"! #as FF$AA (SD I 1$21, in I 2F, ax I F<), and for the )A gro"! it #as A9$9: (SD I A$:F, in I 1A, ax I =1)$ Analysis on the 5AS scales sho#ed that the task did not have an effect on the !artici!ants; ood$ Hn each diension, 2 2 ixed AGH5As #ith tie ('efore or after) as the #ithin&s"'8ects factor and gro"! (*A or )A) as the 'et#een&s"'8ects factor revealed nonsignificant ain effects of tie 23s $2 for anxio"s0 and 3(9,F:) I 9$1, ! M $F for coforta'le4$ 7nteraction effects #ith gro"! #ere also nonsignificant 23s $F for sad4$ Eye @oveent Data *atency of Correct Saccades Single&task 'lock vers"s ixed&task 'lock$ 3ig"re 2 sho#s the ean correct saccade latencies for each of the anti& and !rosaccade tasks in each of the single&task and ixed& task 'locks$ A 2 2 2 ixed AGH5A #ith gro"! (*A or )A individ"als) as the 'et#een&s"'8ects factor and task (anti& or !rosaccade) and 'lock ty!e (single task or ixed task) as #ithin&s"'8ects factors revealed highly significant ain effects of task 23(9,F:) I :0$=A, ! Hf a8or i!ortance #ere the effects involving anxiety, as indicated 'y the three&#ay 'lock ty!e J task J gro"! interaction 23(9,F:) I =$::, ! I $094$ %his interaction indicated that, #hereas the *A individ"als exhi'ited shorter antisaccade latencies in the ixed&task 'lock (@ I 2F7$A sec, SD I ==$:A) than in the single&task 'lock (@ I F07$09 sec, SD I AF$=) 2ean difference I &=:$A0 sec0 !(9:) I =$:2,! $0<4$ 7f #e define s#itch cost as the ean latency difference 'et#een the single&task and the ixedtask 'locks, #e can see in 3ig"re F a s#itch 'enefit in antisaccade latency for *A individ"als, '"t not )A individ"als$ Chen #e exained !rosaccade latencies, there #ere no differences 'et#een the ixed&task 'lock (@ I 20A$FF sec, SD I F1$7=) and the single&task 'lock (@ I 9:7$1< sec, SD I 2<$1A) in the *A gro"! 2t(9:) I 9$2:, I $24, and the sae !attern #as o'served in the )A gro"! (for the ixed&task 'lock, @ I 22A$9= sec, SD I 1A$<<0 for the single&task 'lock, @ I 292$17 sec, SD I F=$0F) 2t(20) I 9$=7, ! M $94$ Dro"! differences #ere exained on s#itch cost in 'oth anti& and !rosaccade !erforance$ %he gro"!s differed on antisaccade s#itch cost 2t(F:) I 2$A7, ! I $094, '"t not on !rosaccade s#itch cost (t S#itch vers"s re!eat trials in the ixed&task 'lock$ 3ig"re 1 sho#s ean correct saccade latencies for each of the and& and !rosaccade tasks on the s#itch and re!eat trials in the ixed&task 'lock$ A 2 2 2 ixed odel AGH5A #ith gro"! as the 'et#een& s"'8ects factor and task (anti or !ro) and trial ty!e (s#itch or re!eat) as #ithin&s"'8ects factors revealed a highly significant ain effect of task 23(9,F:) I A7$9<, ! @ore i!ortant to the !"r!oses of the !resent st"dy is the effect of anxiety on !erforance, as s"ggested 'y the significant trial ty!e task gro"! interaction$ %his interaction sho#ed that #hereas *A individ"als 'ecae significantly faster on antisaccade trials #hen s#itching 'et#een antiand !rosaccade trials (@ I 21<$=1 sec, SD I =1$2) than #hen antisaccade trials #ere re!eated (@ I 27=$7 sec, SD I AF$=0) 2ean difference of &2<$0A sec0 B(9:) I F$02, I $0074, the )A individ"als #ere not affected on antisaccade !erforance 2@ I 2::$A= sec, SD I AA$F:, #hen s#itching 'et#een trials, and @ I 2::$07 sec, SD I A7$:2, #hen antisaccade trials #ere re!eated0 ean difference I 0$1:0 t(20) Dro"! differences #ere exained on s#itch costs for 'oth anti& and !rosaccade !erforance$ %he gro"!s differed on antisaccade s#itch cost 2t(F:) I 2$17, ! I $094, '"t not on !rosaccade s#itch cost 2t(F:) I 9$92, ! I $2<4$ 7t #as !redicted that anxiety #o"ld have a negative i!act on s#itching 'eca"se efficient !erforance on the s#itching tasks de!ends on the availa'ility of #orking eory reso"rces, and since anxiety diinishes availa'le reso"rces, !erforance on s"ch cognitive tasks #o"ld 'e ex!ected to deteriorate$ %he !resented findings collectively s"!!ort this vie#$ +ercentage of Saccade Error Single&task 'lock vers"s ixed&task 'lock$ 3ig"re = de!icts the error rates for each of the *A and )A gro"!s on the anti& and !rosaccade tasks for the single&task and ixed&task 'locks$ %he !ercentage of trials excl"ded fro the analysis did not differ 'et#een the t#o gro"!s 2t(F:) S#itch vers"s re!eat trials in the ixed&task 'lock$ 3ig"re 7 de!icts the error rates for each of the *A and )A gro"!s on the anti& and !rosaccade tasks for the re!eat and s#itch trials$ %he !ercentage of trials excl"ded fro the analysis did not differ 'et#een the t#o gro"!s 2t (F:) /elation (et#een Saccade *atency and Error /ate S#itch Cost Des!ite the nonsignificant task trial ty!e and task trial ty!e gro"! interactions on analysis of error rates, #e decided to cond"ct a f"rther analysis to exaine #hether the shorter antisaccade latencies, as o'served on the s#itch trials, and the higher error rates on these trials co"ld 'e ex!lained in ters of a s!eed&acc"racy trade&off$ Ce exained this relationshi! for each gro"! se!arately, 'y analy.ing !erforance on the single task #ith that on the ixed task and 'y analy.ing this relationshi! in the ixed 'lock only$ %he relationshi! #as not significant for either the *A or the )A gro"! (r I &$99, ! I $=, and r I &$02, ! I $:, res!ectively) #hen #e co!ared !erforance on the single task #ith that on the ixed task$ Siilar effects #ere fo"nd #hen #e exained the relationshi! in the ixed 'lock only (r I &F1, ! I $91,andr I &$92, ! I $=, for the *A and the )A gro"!s, res!ectively)$ %hese res"lts indicate that saccade latency i!roveent #as not si!ly the res"lt of a s!eed&acc"racy trade&off$ (ehavioral Data %he !artici!ants #ere generally slo#er to identify the direction of the target arro# in the ixed&task 'lock (@ I 1<7$:2 sec, SD I 929$<1) than in the single&task 'lock (@ I 1=F$A7 sec, SD I :F$A) and #ere slo#er on the antisaccadetask (@ I 1<2$11 sec, SD I 90<$AA) than on the !rosaccade task (@ I 1=:$0A sec, SD I 907$0A)$ A 2 2 2 ixed AGH5A on ean reaction tie to target arro#, #ith task and 'lock ty!e as #ithin& s"'8ects factors and gro"! as a 'et#een&s"'8ects factor, revealed a significant ain effect of 'lock ty!e 23(9,F:) I 99$12, ! D7SC-SS7HG H"r res"lts s"!!ort the !redicted effects, #ith greater error rates and longer latencies on the antisaccade than on the !rosaccade task$ %he !artici!ants exhi'ited the !redicted s#itch 'enefit in antisaccade latency #hen re6"ired to s#itch 'et#een anti& and !rosaccade trials (ixed&task 'lock), as co!ared #ith #hen they co!leted anti& and !rosaccade tasks se!arately in 'locks of re!eated trials (single&task 'locks)$ %he !aradoxical i!roveent in antisaccade latency is consistent #ith the res"lts of !revio"s research (e$g$, (arton et al$, 20020 Cherkasova et al$, 20020 )odgson et al$, 20010 @anoach et al$, 2002), in #hich siilar !aradoxical red"ctions in antisaccade latencies #ere re!orted #hen !artici!ants #ere re6"ired to s#itch$ %he i!roveent in antisaccade latency o'served in the !resent st"dy can 'e associated #ith the !rec"ing of the task goal on every trial in the ixed 'lock, as co!ared #ith the single&task 'lock (see ,ray, 200=0 Gie"#enh"is et al$, 2001)$ 7n the ixed&task 'lock, !rec"ing facilitated task set reconfig"ration and allo#ed attentional reso"rces to 'e allocated to the c"rrent task, res"lting in shorter antisaccade latencies$ 7n the single&task 'lock, ho#ever, the !artici!ants had to actively aintain and trigger the task goal in #orking eory thro"gho"t each anti& and !rosaccade 'lock, in t"rn res"lting in slo#er task set reconfig"ration and, hence, slo#er generation of a correct antisaccade$ Hf a8or i!ortance to the ais of the !resent ex!erient #ere the findings regarding the effect of anxiety in relation to s#itching 'enefit on correct antisaccade latencies$ Co!aring antisaccade !erforance in the singletask and the ixed&task 'locks sho#ed that *A individ"als exhi'ited the ex!ected s#itch 'enefit in correct antisaccade latencies '"t that )A individ"als sho#ed no significant i!roveent$ )o#ever, the t#o gro"!s did not differ in ters of their !erforance on !rosaccade trials$ %he i!ortant gro"! difference !oints to a diinished a'ility )A individ"als to "tili.e the c"e to facilitate task set reconfig"ration$ Cithin the frae#ork of the attentional control theory, these findings s"ggest that these individ"als are less a'le to exercise to!&do#n attentional control to efficiently shift attentional reso"rces according to ne# task deands$ %his fail"re to "se to!&do#n control efficiently can 'e attri'"ted to diinished #orking eory ca!acity in )A individ"als$ 7ndeed, !revio"s research e!hasi.es the critical role of #orking eory ca!acity in task set reconfig"ration, #hich deterines s"ccessf"l task&s#itching !erforance (e$g$, (addeley et al$, 20090 )ester & Daravan, 200A0 ,ane et al$, 20090 *ogan, 2001)$ )o#ever, in order to ens"re the validity of the concl"sions 'ased on these findings, it #as cr"cial to disentangle the !ossi'le contri'"tion of !rec"ing (in the ixed&task 'lock) fro that of task s#itching in anti& and !rosaccade !erforance$ 7n other #ords, it is "nclear #hether the o'served !erforance i!roveent is a tr"e reflection of the effect of anxiety on task s#itching in this !aradig or si!ly the res"lt of a confo"nd associated #ith c"ing task goal in the ixed&task, '"t not in the single&task, 'lock$ %o do this, trial&'y&trial analyses #ere cond"cted co!aring !erforance on re!eat and s#itch trials in the ixed&task 'lock$ %he res"lts re!licated the !aradoxical i!roveent in antisaccade latencies o'served #hen !erforance in the single&task 'lock #as co!ared #ith that in the ixed&task 'lock$ @ore s!ecifically, the !artici!ants exhi'ited a s#itch 'enefit in antisaccade latencies #ithin the ixed&task 'lock #hen the antisaccade trial #as !receded 'y a !rosaccade trial (s#itch trial), as co!ared #ith #hen it #as !receded 'y a trial of the sae task ty!e (re!eat trial)$ Cr"cially, this i!roveent #as fo"nd to 'e associated solely #ith the antisaccade !erforance of *A individ"als, #hereas )A individ"als exhi'ited no s"ch i!roveent$ %hese findings have i!ortant i!lications for "nderstanding the "nderlying echaniss 'y #hich anxiety affects task& s#itching !erforance$ According to the attentional control theory (Eysenck et al$, 2007), s"ccessf"l s#itching !erforance re6"ires that attentional control 'e exercised in a !ositive #ay to allocate attentional reso"rces to the task relevant to the c"rrent goal$ 7n the ixed&task 'lock, s#itching fro a !rosaccade task to an antisaccade task facilitated antisaccade !erforance in the *A individ"als, '"t not in the )A individ"als, des!ite the ex!licit !resentation of the task goal on every trial$ Chen !rosaccade !erforance is exained, on the other hand, *A individ"als are "ch slo#er #hen re6"ired to s#itch to a !rosaccade task than #hen !rosaccades are re!eated$ %he )A gro"! #as again "naffected 'y s#itching on these trials$ %he key to "nderstanding these gro"! differences ay lie in an "nderstanding of the effect of !erforing an attentionally deanding cognitive task (s"ch as task s#itching) in con8"nction #ith antisaccade !erforance$ A recent st"dy in #hich !recisely this relationshi! #as looked at (,rist8ansson, Chen, & Gakayaa, 2009) sho#ed that antisaccade !erforance i!roved significantly #hen it #as !erfored si"ltaneo"sly #ith a deanding !erce!t"al task$ %he a"thors reasoned that the si"ltaneo"s !erforance of other attention&re6"iring tasks interferes #ith the !rograing of a reflexive res!onse, 'oth facilitating antisaccade !erforance and delaying !rosaccade !erforance$ %hey concl"ded that in the antisaccade task, ?less attention is ore$? 7n light of these findings and the findings of the !resent st"dy, it is therefore conceiva'le that task s#itching facilitates antisaccade !erforance '"t diinishes !rosaccade !erforance$ +roviding s"!!ort for the ass"!tions of the attentional control theory, die !resent findings s"ggest that )A individ"als are "na'le to exercise efficient to!&do#n attentional control to distri'"te attentional reso"rces according to task deands$ %his is in line #ith ne"ro'iological evidence sho#ing decreased activation in the D*3+C and the 5*3+C, 'elieved to 'e involved in od"lation of to!&do#n attentional control in )A individ"als$ (isho! (2007) re!orted that high levels of anxiety #ere associated #ith red"ced recr"itent of these areas and arg"ed that this is indicative of a diinished a'ility in )A individ"als to recr"it these areas to a"gent attentional control$ Another i!ortant o'servation in the !resent st"dy #as the lack of anxiety&related effects on saccade acc"racy (on 'oth single&task vs$ ixed&task and re!eat vs$ s#itch trials #ithin the ixed&task 'lock), des!ite significant effects on correct antisaccade latency$ %his finding is f"lly consistent #ith the !rediction of the attentional control theory that anxiety is ore likely to i!air !erforance efficiency than !erforance effectiveness$ %he lack of anxiety&related differential effects on antisaccade acc"racy is consistent #ith !revio"s findings (Derakshan et al$, in !ress0 Derakshan & Eysenck, 9::<0 Ellian, Dreen, /ogers, & 3inch, 9::7)$ 3or exa!le, Ellian et al$ fo"nd no effect of anxiety on the n"'er of correct res!onses on a s"stained attention task$ )o#ever, )A individ"als took longer to res!ond, #hich #as attri'"ted to greater effort 'eing exerted in !erforing the task and, hence, decreased !erforance efficiency$ As a final !oint, in this article, #e have ass"ed that correct antisaccade !erforance reflects the individ"al !artici!ant;s level of attentional control (e$g$, Ettinger et al$, 200<)$ )o#ever, an alternative ass"!tion is that the control of eye oveents in the antisaccade task refleets soe for of 'ehavioral inhi'ition that ay or ay not 'e linked #ith attention$ A third !ossi'ility is that there is an interaction 'et#een attentional control and 'ehavioral inhi'ition in #hich attentional control facilitates 'ehavioral inhi'ition$ Diven the #ealth of data on the ne"ral correlates of antisaccade !erforance (see Ettinger et al$, 200<), #e arg"e that saccadic eye oveents in the antisaccade task are relia'le indices of attentional control$ Ce e!hasi.e the i!ortance of attentional control in "nderstanding antisaccade !erforance and, at the sae tie, acce!t the need for ore research in identifying the role of 'ehavioral inhi'ition$ Hverall, the findings of the !resent ex!erient sho# that anxiety diinishes the coonly exhi'ited s#itching 'enefit in correct antisaccade latency, s"ggesting that #hen shifting, )A individ"als cannot i!leent goaldirected to!&do#n control to allocate attentional reso"rces to the c"rrent task deands$ 3"rtherore, anxiety did not have an effect on directional acc"racy$ %hese o'servations are !recisely in line #ith the !redictions of the attentional control theory (Eysenck et al$, 2007), !roviding a ore direct assessent of the "nderlying echaniss #ith #hich anxiety i!airs attentional control #hen the shifting f"nction is re6"ired$ %he !resent findings indicate that anxiety i!airs efficient shifting of attentional reso"rces to task deands in the a'sence of threat$ %heoretically, it is i!ortant to exaine ho# threat affects task&s#itching !erforance in )A and *A individ"als$ %here is "ch research in s"!!ort of the association 'et#een anxiety and increased attentional 'ias to#ard threatening sti"li (e$g$, 3ox, /"sso, (o#les, & D"tton, 2009 0 3ox, /"sso, & D"tton, 20020 see (ar&)ai et al$, 2007, for a revie#)$ %he attentional control theory !redicts that the adverse effect of anxiety on task !erforance #ill 'e greater in the !resence of task&irrelevant threat&related aterial$ Ce are c"rrently investigating attentional control in the !resence of eotional sti"li$ 3inally, it is i!ortant to investigate attentional control as eas"red thro"gh the antisaccade task in other ne"ro!sychiatric conditions, s"ch as !sychosis and de!ression$ Diven the high correlation 'et#een anxiety and de!ression, it #ill 'e "sef"l for f"t"re research to atte!t to disentangle the "ni6"e contri'"tions of these conditions in ters of attentional control$ H"r la' is c"rrently investigating the echaniss "nderlying attentional control #ith res!ect to de!ression$ GH%E 9$ %he ain !"r!ose that the arro# identification task served #as to ens"re the !artici!ants; engageent in the ain task$