Perception & Psychophysics
2000,62 (7), 1367-1382
A central-peripheral asymmetryin maskedpriming
FRIEDERIKE SCHI,AGHECKENANdMARTIN EIMER
Birkbeck College, London, England,
Masked primes presented prior to a target result in behavioral benefits on incompatible trials (in
which the prime and the target ile mapped onto opposite responses) when they appear at fixation, but
in behavioral benefits on compatible trials (in which the prime and the taxget are mapped onto the
same response) when appearing peripherally. In Experiment 1, the time course of this centra'Iperipherat asymmetry (CPA) was investigated. For central primes, compatible-trial benefits at short
stimulus onset asynchronies(SOAs) turned into incompatible-trial benefits at longer SOAs.For peripheral primes, compatible-trial benefits at short SOAs increased in size with longer SOAs. Experiment 2 showed that these effects also occur when primes and targets are physically dissimilar, ruling
out an interpretation in terms of the perceptual properties of the stimulus material. In Experiments 3
and 4, the question was investigated as to whether the CPA is related to visual-spatial attention and./or
retinal eccentricity per se. The results indicate that the CPA is independent of attentional factors but
strongly related to the physiological inhomogeneity of the retina. It is argued that central and peripheral primes trigger an initial motor activation, which is inhibited only if primes are presented at retinal
locations ofsufficiently high perceptual sensitivity.The results are discussedin terms ofan activation
threshold model.
In the last decade,the role of inhibition in cognitive processeshas receivedconsiderableattention. On the basis
of evidence from different experimentalparadigms,the
concept of inhibitory control of information processing
is gaining increasing support. Possibleexamples of inhibitory control currently under discussioninclude inhibitory processingin the perceptionand/or encodingofrepeated items (repetition blindness; Kanwisher, 1987); an
inhibitory bias in spatial attention (inhibition ofreturn;
Mayloq 1985),the inhibition of visual processingas a result of target identification (attentional blink, Raymond,
Shapiro,& Arnell, 1992),theinhibition of irrelevantobjects during target selection (negativepriming; Tipper,
1985;seeFox, 1995,and May, Kane, & Hasher,1995,for
reviews), and the selectiveinhibition of ongoing motor
activity (inhibitory motor control; De Jong, Coles, &
Logan, 1995; Kopp, Rist, & Mattler, 1996).Although the
specific mechanismsunderlying these phenomenamay
differ, they are likely to reflect the operation ofadaptively
significant inhibitory control processes,rather than just
basic capacity limitations ofthe cognitive system (seeArbuthnott, 1995,and Houghton& Tipper, 1996,for further
discussions).
An area in which inhibitory effects generally have not
beenfound is the field of perceptionwithout subjective
awareness,or near-thresholdperception (for a notable exThe present research was supported by grants from the Biotechnology and Biological Sciences ResearchCouncil and from the Deutsche
Forschungsgemeinschaft. The authors thank Arno von Buxhoeveden
and Jutta Braune for their help in running the experiments. Correspondence concerning this article should be addressedto F. Schlaghecken,
Department of Experimental Psychology, Birkbeck College, Malet
Street,London WClE, 7HX, England (e-mail: t'.schlaghecken@bbk.
ac.uk).
ception,seeDagenbach,Carr, & Wilhelmsen, 1989).For
example, inhibiting the attentional shift toward the location of a highly invalid peripheral cue is possible with
suprathresholdcues,but impossibleifcues are presented
near detectionthreshold (McCormick, 1997). Similarly,
inhibiting one of two meaningsof a polysemousword is
possible for suprathreshold words, but not for nearthreshold words (Marcel, 1980). Finally, negativepriming effects have been found to turn into positive priming
effects when probe-trial distractors were successfully
masked(Allport, Tipper, & Chmiel, 1985; seealso Neill,
Valdes,& Terry, 1995).
Recently, however, evidence has been reported that
even under conditions of near-threshold stimulus presentation,inhibitory processescan be observed(Eimer,
1999; Eimer & Schlaghecken,1998; Schlaghecken&
Eimer, 1997). In the maskedprime paradigm,participants had to perform a choice reaction time (RT) task in
responseto simple visual stimuli, such as arrows or geometric figures, that were precededby maskedprimes.l In
compatible trials, the prime and the target were identical;
in incompatible trials, they were mapped to different responses;in neutral trials, the prime was task irrelevant.
Surprisingly, performance benefits (fast responses,low
error rates) were obtained in lncompatible trials, whereas
performancecostswere presentin compatibletrials.
First insights into the basis of this unexpectednegative compatibility effect were obtained by inspecting the
lateralizedreadinesspotential (LRP), an electrophysiological measureof unimanual responseactivation (Eimer,
1999; Eimer & Schlaghecken,1998): Around 200 msec
after prime onset, the LRP showed a partial activation of
the responseassignedto the prime stimulus. However,
about 100 msec later, in the time range in which responses
1367
Copyright 2000 PsychonomicSociety,Inc.
1368
SCHLAGHECKENAND EIMER
to the target stimulus were selectedand activated,this
initial responseactivation was replaced by an activation
ofthe oppositeresponse.It was arguedthat the early partial activation of the responseassignedto the prime is a
consequenceof direct perceptuomotor links, whereasthe
subsequentreversalphasereflects an active inhibition of
this initial responsetendency.The negative compatibility
effect on behavioral performance would thus reflect an
inhibition of a responsetendencytriggered by the masked
primes. This activation-followed-by-inhibition hypothesis can also account for the fact that performance benefits for compatible trials (positive compatibitity effects)
were observed in experiments in which the masking
stimulus itself servedas the target(Klotz & Wolff, 1995;
Neumann & Klotz, 1994): When the target immediately
follows the prime instead of being separatedfrom thL
prime by a masking stimulus, prime-target stimulus onset
asynchronies(SOAs) are shorteE so that the activation
of the responseto the target is more likely to overlap with
the initial, prime-relatedpartial responseactivation.This
w.ouldresult in performancebenefits in compatible trials, owing to the partial activation ofthe correci response,
and in costs in incompatible trials, owing to the partial
activation of the incorrect response.
In accordancewith the activation-followed-by-inhibition
account,experimentsthat varied the SOA betweenmask
and target(Eimer, 1999,Experiment 3; Schlaghecken&
Eimer, 1997) found positive compatibility effects for a
mask*targetSOA of 0 msec,but theseeffectsdecreased
with longer SOAs and eventually turned into negative
compatibility effects. Moreover, negativecompatibility
effects were also obtained when responseswith two fingers of the same hand were required and even when only
a single overt responsewas to be executedin a go/no-go
experiment(Eimer & Schlaghecken,1998,Experiment 3):
When go stimuli were deliveredas primes, responsesto
go targets were delayed and false alarms to no-go stimuli
were less frequent, relative to trials in which a no-so
stimulus servedas the prime, suggestingthat the responie
assignedto the go prime was actively inhibited.
In the studies described above, behavioral and electrophysiological evidence for activation-followed-byinhibition was found when maskedprimes were presented
at fixation. However. with masked primes preiented in
the periphery, a very different pattern of results emerged.
Schlagheckenand Eimer (1997) compared the behavioral effectsof foveally and peripherally presentedmasked
primes and found that with a 96-msecmask-targetSOA,
central primes elicited negative compatibility effects,
w-hereasperipheral primes elicitedpositive compatibility
effects(i.e., performancebenefits for compatibletrialsj.
How does this central-peripheral asymmetry (CpA)
fit into the activation-followed-by-inhibition framework?
A comparablesituation, in which the influence of (unmasked)distractor stimuli on responsesto targetsvaries
with the spatial separationoftarget and distractors,can
be found intheflanker compatibilityparadigm(e.g.,B. A.
Eriksen & C. W. Eriksen, 1974): Similar to the masked
prime paradigm,psychophysiologicalevidenceindicates
that this interference is due to the fact that distractors
that are potential target stimuli are likely to trigger a partial activation of their corresponding motor response
(Gratton, Coles, Sirevaag,Eriksen, & Donchin, 1988;
Smid, Mulder, & Mulder, 1990). In addition. similar to
the masked prime paradigm, the interference effect
changes with changing retinal eccentricity of the distractors (e.g., C. W. Eriksen & Hoffrnann, 1973; C. W.
Eriksen,Pan,& Botella, 1993;Goolkasian,1997;Miller,
1991). However, in the masked prime paradigm, increasedretinal eccentricity ofthe distractorsresultedin
a reversal of the negative compatibility effect, whereas
in the flanker compatibility paradigm, only a gradual decreasein interference effects has been observed, without
evidencefor "a different kind of processing" (Goolkasian, 1997) of central and peripheral information.
The present experimentswere conducted to investigate whether the CPA in prime-target compatibility effects reflects qualitatively different types of processing
for central and peripheral stimuli. If it could be established that there are systematic differences in the way
that foveal information and peripheral information affect
motor processes,this may have implications for our understandingof the mechanismsmediating betweenperceptionand action and ofthe role ofinhibitory processes
in perceptuomotorlinks. Experiment I was conductedto
confirm that the CPA is a reliable effect in maskedpriming situations and to investigatethe time course of the
effects of central and peripheral primes on behavioral
performance. Experiment 2 was conductedto rule out the
idea that the CPA is primarily basedon perceptualproperties of the stimuli employedpreviously,ratherthan reflecting different patternsofresponse activationand inhibition. Experiments3 and4 studiedthe relation between
spatial attention and the CPA, as well as the role of physiological propertiesof the visual systemcorrelatedwith
retinal eccentricity.
EXPERIMENT
I
Experiment I was conductedto test whether primetarget compatibility effects elicited by central and peripheral masked primes reflect the same underlying piocesses,which vary only in their latencies, or whether
they reflect qualitatively different processes.The onset
of the partial responseactivation triggered by the prime
may simply be delayed for peripheral relative to foveally
presentedprimes-for example, becauseperceptual
analysis takes longer for peripherally presentedinformation than for centrally presentedinformation (seeCoyle,
1994).If this were the case,then with peripheralprimes,
the activation ofthe responserequired by the targetwould
be more likely to overlap with the initial, prime-related
activation, insteadofoccurring during the subsequentresponseinhibition phase.In Experiment lA, prime-target
compatibility effects for central and peripheral primes
were investigatedwith short to intermediate mask-tarset
CENTRAL-PERIPHERALASYMMETRYIN MASKED PRIMING
SOAs (G-96 msec). If the CPA is due to a delayed response
activation triggered by peripheral primes, then with a
0-msec SOA, no compatibility effects should be elicited
by peripheral primes while central primes should give
rise to a positive compatibility effect. Increasing the SOA
should result in gradually emerging positive compatibility
effects for peripheral primes and in a reversal from positive to negative compatibility effects with central primes.2
Alternatively, the onset of the inhibition phase may be
delayed with peripheral primes, causing prime-target
compatibility effects to reverse polarity later for peripherally than for centrally presented primes. In Experiment lB, prime-target compatibility effects were investigated with intermediate to long SOAs (9G-192 msec) in
order to test whether peripherally presented primes would
give rise to negative compatibility
effects at longer
SOAs. Moreover, the possibility that the negative compatibility effect elicited by central primes would remain
stable with longer mask-target SOAs was tested. If, however, positive compatibility effects were observed for peripheral primes at all SOAs, this would indicate that inhibitory processes may be entirely restricted to response
tendencies elicited by foveally presented information,
whereas response activations triggered by peripheral information may not be actively inhibited at all.
Method
Participants. Twelvepaid volunteers(5 males),21 37 yearsof
age (mean age,29.l years),participatedin Experiment lA, and
l2 paid volunteers(6 males),l8-38 yearsof age (mean age,
28.7 years),participatedin Experiment lB. All the participants
were right-handedand had normal or corrected-to-normalvision.
Stimuli and Apparatus. Left- andright-pointing doublearrows
(<< and >>) servedasprime stimuli. The maskwas constructedby
superimposingthesestimuli uponone another.The targetconsisted
of two left- or right-pointingdoublearrows(<< << and >> >>),
spacedappropriatelyto allow the prime and the mask to fit exactly
into the central gap. All the stimuli were presentedin black on a
white backgroundon a l7-in. computerscreen.Prime and mask
stimulisubtended
a visualangleof approximatelyl. l5' X 0.4o;target stimuli subtendeda visual angleof approximately3.45' X 0.4".
Procedure. The participantswere seatedin a dimly lit, soundattenuatedchamber,with responsebuttonsundertheir left andright
index fingers. A computerscreenwasplaced 100cm in front ofthe
participants' eyesso that the screencenterwas in the centerofthe
participants'horizontalstraightaheadline of sight.The participants
wereinstructedto maintaincentraleyefixation andto respondasfast
and accuratelyas possiblewith a left buttonpressto a left-pointing
targetarrow and with a right buttonpressto a right-pointing target
Experimentalblocks consistedof80 trials, eachstartingwith the
presentationof a prime stimulus(16-msecduration),followed by a
mask (100-msecduration)and a targetstimulus(100-msecduration). In the central condition,maskedprimes appearedat fixation.
lnthe peripheral condition, maskedprimesappearedrandomlyand
with equalprobability either2.8oaboveor 2.8obelow fixation. Target stimuli appearedalways to the left and right of fixation (see
above).Trials were termed compatiblewhen prime and target arrows were pointing in the samedirection, and incompatible otherwise. Both conditionswere equiprobableand randomizedwithin a
givenblock.
1369
The maskalwaysfollowedthe prime immediately,whereasmasktargetSOA was blockedand was 0,32, 64, or 96 msec (Experiment lA) or 96, 128, 160,or 192 msec(Experiment1B). The intertrial interval was I sec. Different prime locations (central and
peripheral) were also blocked. Since each ofthe four SOAs was
combinedwith eachof the two prime locations,therewas a total of
eight differentlocation/SOAcombinationsin eachexperiment.Each
location/SOAcombinationwas deliveredin four blocks, resulting
in a total of 32 experimentalblocks per experiment.The experiments were divided into four parts, each containing one block of
eachlocation/SOAcombination.Within eachpart,the orderin which
theseblocks were deliveredwas randomized.3
Data analysis. Repeatedmeasuresanalysesofvariance (ANOVAs)
were performed on RTs and error ratesin the experimentalblocks
for the factorsofprime position (centralor peripheral),SOA (0, 32,
64, or 96 msecin Experiment 1A and 96, 128, 160,or 192msecin
Experiment1B),compatibility(compatibleor incompatible),andresponseside(left or right). Whereappropriate,Greenhouse-Geisser
adjustmentsto the degreesoffreedom wereperformed(indicatedin
the Resultssectionby e).
Results
In both experiments,main effects of prime position
were obtained for RTs and error rat;s [all ,F's(I , I I ) > 8.7,
all ps < .013], since RTs were shorter and fewer errors
were made in the peripheralprime condition (372 msec,
L9oloerrors in Experiment 1A, and 358 msec, I .9o%in Experiment lB) than in the centralprime condition (396 msec,
2.7oh,and 377 msec, 2.9o/o,respectively).
Experiment 1A. Overall, RTs were shorter and fewer
errors were made in compatible trials than in incompatibletrials[377msec,1.8%vs. 391msec,2.9%;F(l,ll) :
63.04 and 31.59,bothps < .001,for RT and error rates,
respectively]. For RT, this effect was larger for peripheral
p r i m e s t h a n f o r c e n t r a lp r i m e s [ F ( 1 , 1 1 ) : l l . 9 8 , p <
.005], and for both RTs and error ratespositive compatibility effects were more pronounced with short SOAs
thanwith long SOAs[RT: F(3,33) : 19.20,p<.001,e :
.694; error rates:F(3,33): ll.27,p < .001, e: .8271.
As can be seem from Figure 1, the positive compatibility effect not only becamesmaller with increasingSOA
for central primes, but turned into a negative compatibility effect with the 96-msecSOA. In contrast,positive
compatibility effects became larger with increasing
SOAs for peripheral primes. This pattern was reflected
inprimeposition x SOA x compatibility interactionsfor
RTs and error rates[F(3,33) : 3 5.07 and I 8.5, bothps <
.001,e: .590 and.669,respectivelyl.
Pairwisecomparisonsof RTs for compatibleand incompatible trials, conducted with two-tailed I tests for individual SOAs and prime positions, confirmed that for peripheral primes, RT for compatible trials was sigaificantly
shorter than RT for incompatible trials in all SOA conditions [all rs(l l) > 5.9, allps < .001]. For centralprimes,
RT for compatible trials was significantly shorter than
RT for incompatible trials in the 0-msec SOA condition
and in the 32-msecSOA condition [both rs(ll) > 6.4,
bothps < .0011,whereas it was significantly longer in
the 96-msecSOA condition[r(1 l) : 4.06,p < .002].No
l 370
SCHLAGHECKEN
AND EIMER
comp.
{F
fl-
incomp.
a
E
F
tr
-r-t
-r--t L--J---1
E-n-il
1
o
Mask-Target
SOA(ms)
Figure l. Experiment 1: Reaction times (RTs; line graphs) and error rates (bar graphs) in compatible trials(gray) and incompatible trials (white) for the four stimulus onset asynchrony (SOA) condiiions
in Experiment lA (0,32, 64,and,96 msec) and Experiment lB (96, 128, 160, and 192 msec), displayed
separately for central primes (left) and peripheral primes (right).
significant compatibility effect was present for central
primesin the 64-msecSOA condition.
Experiment lB. No main effect of compatibility was
present,but highly significant compatibility X prime
position interactionsfor RTs and error rates [F( l,l I ) :
31.82and 19.66,respectively;bothps < .0011indicated
the presenceof prime-target compatibility effectsof opposite directions for the two prime positions (see Figure 1). With peripheralprimes, a stablepositive compatibility effect was found, whereaswith central Drimes.a
stablenegativecompatibility effect was observed.Additional ANOVAs, conducted separatelyfor central trials
and peripheral trials, confirmed that these effecrs were
significantfor peripheralprimes [RT: F(1,11) :73.20,
p < . 0 0 1 ;e r r o rr a t e :f ( l , l l ) : l 5 . l 0 , p < . 0 0 3 1a, s w e l l
a s f o r c e n t r a lp r i m e s [ R T : F ( 1 , 1 l ) : 1 4 . 1 3 p
, < .003;
error rate:f'(l,l l) : 8.42,p < .0l4l.Importantly, there
was no compatibility X SOA interactionon either RT or
error rates for either central or peripheral primes, indicating that the respectivepriming effects were of equivalentsizeacrossall SOAs.
Discussion of Experiment I
The effect of maskedprime stimuli on performanceto
subsequentlypresentedtargetsdependscrucially on the
location of these primes. With mask-target SOAs of
96 msec, negative compatibility effects were observed
with centrally presentedprimes, but positive compatibility effects were present for peripheral primes (Schlaghecken& Eimer, 1997).Experiment I was conductedto
test whether this CPA was due to latency differencesin
the responseactivation and/or inhibition processestriggeredby central and peripheralprimes. To this purpose,
the time course of priming effects was investigatedfor
central and peripheralmaskedprimes by varying masktarget SOA between0 and 192 msec.
The results confirmed that the CPA is a reliable phenomenonthat can be obtainedunder a number of different SOA conditions.None of theseconditions,however,
revealedany evidencethat the CPA is due to latency differences:In Experiment lA, peripheralprimes produced
a substantialpositive compatibility effect in the O-msec
SOA condition, indicating that the onsetofresponseten-
CENTRAL-PERIPHERALASYMMETRYIN MASKED PRIMING
dencies triggered by peripheral primes was not (or not
substantially)delayed,as comparedwith responsetendencies triggeredby centralprimes. In Experiment lB, both
positive compatibility effects obtained with peripheral
primes and negative compatibility effects obtained with
central primes remained constant acrossthe whole range
of SOAs employed in this study. This suggeststhat response inhibition elicited in the central prime condition
is not a transient phenomenon that is subject to rapid
decay,whereasresponseinhibition is entirely absentin
the peripheral prime condition. Although responseinhibition with peripheral primes may have been obtained at
even longer SOAs, the fact that the positive compatibility
effects showed no sign of decreasewith increasing SOA
makes this assumption unlikely.
However, it has to be noted that the positive compatibility effect obtained for the 0-msec SOA was significantly smaller for peripheralthan for central primes [12
vs. 33 msec;(l l) : 4.47,p<.0011. This differencemay
indicate that during the responseactivation for a O-msec
SOA target, the responsetendency triggered by central
primes was already fully activated, whereas it was only
just beginning to build up for peripheral primes. If this
were the case,positive compatibility effects in response
to peripheralprimes in the 0-msecSOA condition should
be restrictedto relatively slow responses,becausefaster
responseswould be selectedand activatedprior to the onset of the prime-related responseactivation. This was
tested by comparing RT effects for peripheral primes in
the O-msecSOA condition for the fastest l0% of all responses(mean RT of about 300 msec) with the effects
obtained for the slowest l0% of all responses(mean RT
of about 415 msec), obtained on the basis of individual
RT distributions. Positivecompatibility effectswere obtainedfor slow responses[( I 1) : 2.97,p < .0I 3], aswell
as for fast responses[/(l l) : 4.62, p < .0011,and these
effectsdid not differ in size [r(l l) < 1.3,n.s.]. Assuming
that the selection and activation of the responseto the
target precedesexecutionby at least 100 msec, the fact
that positive compatibility effects were present for the
fastestresponsesin the 0-msecSOA condition indicates
that peripheral primes affected motor activation around
200 msec after their onset.This estimatecorrespondswell
with LRP evidenceobtainedwith central primes (Eimer
& Schlaghecken,1998), demonstratingthat the partial
activation of the primed responsestartsabout 200 msec
after prime onset,and suggeststhat the onsetof the primed
responseactivation is not systematicallydelayedfor peripheral relative to central primes.
Taken together, the results of Experiment I support
the idea that there are qualitative differences in response
activation processestriggered by central and peripheral
primes: Whereas central primes give rise to an activationfollowed-by-inhibition process,peripheral primes seem
to elicit a processofactivation-only. Since this conclusion
may have important implications for our understanding
of how visual information affects performance, alternative interpretationsand possibleconfounding factors have
to be carefully considered.In Experiment2,the question
1371
ofwhether the CPA is based on perceptual differences
that are due to stimulus presentation was investigated: Perhapsthe rapid successionofcentral primes and targetshas
resulted in specific emergent features that did not occur
with peripheral primes. Experiments 3 and 4 addressed
the question of attentional differences: Because all the
targets were presented at fixation, central primes were
processedwithin the focus of attention, whereas peripheral primes, presentedat unattended locations, were not.
EXPERIMENT
2
In Experiment I, prime and target arrows were identical in compatible trials, while pointing in opposite directions in incompatibletrials. The successivepresentation
ofprimes and targets at adjacent locations in the central
presentation condition may have resulted in qualitatively
different emergent visual features in compatible and incompatible trials, resulting in an advantage for incompatible trials in the central presentationcondition. Owing
to the spatial separation between primes and targets, no
such emergent features would be elicited in the peripheral presentation condition. In addition, the presentation
of two identical stimuli in rapid successionand close
spatial proximity may impair the detection of the second
stimulus. relative to a situation in which these stimuli are
not identical or appear at different locations Qterceptual
repetition blindness; seeHochhaus & Johnston, 1996). If
this phenomenon was responsible for the negative compatibility effects obtained with central primes and longer
SOAs in Experiment l, this may fully explain the presence
of a CPA. To investigatetheseissues,physically similar,
as well as physically dissimilar, prime and target stimuli
were employed in Experiment 2. If the CPA was basedon
perceptual properties of the stimuli employed in Experiment I, it should be obtained inthe similar condition but
should disappear inthe dissimilar condition.
Method
(7 female),l9-39 yearsof
Participants.Eightpaidvolunteers
All the
in theexperiment.
age(meanage,26.5years),participated
andhadnormalor corrected-toparticipantswereright-handed
normalvision.
Stimuli, Apparatus,and Procedure.The stimuli,apparatus,
SOAconditions
wereidenticalto the0- and128-msec
andprocedure
exceptthatdifferenttarget
I A andI B, respectively,
ofExperiments
In 50%ofall thetrials,centralarrowtargets
stimuliwereemployed.
whereas
in theotherhalf,lateraltargetstimuliwere
werepresented,
oftwo left-pointingor rightdelivered.Centraltargetsconsisted
pointingdoublearrows,appearing
bilaterallyl.5oaboveandbelow
appropriately
fromcenterto center)andspaced
fixationa(measured
asto to allowcentralprimesandmasksto fit exactlyinto thecena
tral gap.Lateraltargetsweresingleplus signs(+), subtending
unilaterally
0.3oX 0.3othatappeared
visualangleof approximately
of 2.4"to theleft or rightof fixation.Cenat a horizontaldistance
order.5
in randomized
tral andlateraltargetswerepresented
to respondwith a left keypress
wereinstructed
Theparticipants
to left-pointingcentraltargetarrowsandto lateraltargetsappearcento right-pointing
ing at theleft sideandwith a rightkeypress
at theright side.
tral targetarrowsandto lateraltargetsappearing
Triafsweretermedcompatiblewhenprimeandtargetarrowswere
appeared
atthe
pointingin thesamedirectionor whena plus-target
1372
SCHLAGHECKEN
AND EIMER
sideindicatedby the prime and incompatibleotherwise.Twentyexperimentalblocksconsistingof 80 trials eachwere run. prime location(centralvs. peripheral)andmask targetSOA (0 vs. I 28 msec)
wereblocked,with five successive
blocksfor eachofthe four combinationsoftheseconditions.At the beginningofeachofthesefour
experimentalsegments,
a briefpracticeblock was delivered.
Data analysis. RepeatedmeasuresANOVAs wereperformedon
meanRTsand errorrates,separately
for centraland literal targets,
for the factorsofprime position(centralor peripheral),SOA (0 or
128msec),compatibility(compatible
or incompatible),
andresponse
side(left or right). Follow-upanalyseswereincludedto investigate
interactionsbetweenthesefactors.
.016and .024,for centraland lateraltargets,respectively;
error rates; Fs(|,7) : 16.45 and 19.44,ps < .006 and
.003]. However,theseeffectswere accompaniedby prime
position X compatibilityinteractions[RT: Fs( I ,7) : 39.4
and 13.52,ps < .00I and .008; error rates,centraltargets:
F ( I , 7 ) : 2 6 . 2 7 , p< . 0 0 1 ;l a t e r a lt a r g e t sF: ( 1 , 7 ) : 2 . 3 3 ,
n.s.], since positive compatibility effects were much
larger on peripheral prime trials than on central prime
trials. Most important, they were accompaniedby threeway prime position X SOA X compatibility interactions
[ R T :F s ( 1 , 7 ): l l l . 6 7 a n d I l . 0 5 , p s < . 0 0 1 a n d . 0 l 3 , f o r
central and lateral targets, respectively; error rates:
Results
Fs( I,7) : 7 .65 and 7.30,ps < .028 and .03I l: For peFigure 2 showsthe mean RTs and error ratesobtained ripheral primes, positive compatibility effects increased
in compatibleand incompatibletrials in the 0- and 128- in size with increasingSOA, whereasfor central primes,
msec mask-target SOA conditions with central and pe- positive compatibility effectsturned into negativecomripheral primes and central and lateral targets,respectively. patibility effects.
Overall, responseswere faster to lateraltargetsthan to cenFollow-up analysesrevealedthat the main effects of
tral targetsand, similar to Experiment l, fasteron periph- compatibility at the 0-msec SOA were significant for
eral prime trials than on central prime trials.
c e n t r atl a r g e t s[ R T :F ( 1 , 7 ) : 2 9 . 8 3 , p < . 0 0 1 ; e r r o r r a t e s :
For both target types, responseswere generally faster F(|,7) : 12.92,p < .0091and approachedsignificance
and fewer errors were made on compatibletrials than on f o r l a t e r a lt a r g e t s[ R T : F ( 1 , 7 ) : 5 . 2 8 , p < . 0 5 5 ; e r r o r
incompatibletrials [RT: Fs(1,7) : 10.09and 8.29,ps < r a t e s :F ( 1 , 7 ) : 5 . 2 5 ,p < . 0 5 6 1 .T h e p r i m e p o s i t i o nX
centralprimes
400
centraltargets
peripheral
primes
lateraltargets
centraltargets
lateraltargets
380
--ft
incomp
360
a
g
n\
F
E
\
TI
\
340
lf
n
320
300
20
1 0 q
rfl t-h
_--l
0
28
0
o
0
0
128
Mask-Target
SOA(ms)
128
s
128
Figure 2. Experiment 2: Reaction times (RTs; line graphs) and error rates (bar graphs) in compatible trials (gray) and incompatible trials (white) in response to central and lateral targets for the O-msec
mask-target stimulus onset asynchrony (SOA) and the 128-msec mask-target SOA, displayed separately for central primes (left) and peripheral primes (right).
CENTRAL-PERIPHERALASYMMETRY IN MASKED PRIMING
1373
sponseswere selectedand activated during the initial facilitatory phaseand that only slower responseswere subject to responseinhibition.
This was testedby an analysisof RT distributions (20th,
40th, 60th, and 80th percentiles), computed for eachparticipant individually, for the central-prime/128-msecSOA
condition for lateral targets. A responselatency x compatibility
interactionwas obtained lF(3,21) : 6.75,p <
2
Discussion of Experiment
Experiment 2 testedwhether the CPA is causedby per- .013; e : .5711,and further analysesshowedsignificant
ceptual properties of the stimulus material (e.g., emer- negative compatibility effects for the 60th and 80th percentiles [both ts(7) > 2.67, bothps < .032], whereasthe
gent visual features and/or perceptual repetition blindtoward a positive compatibility effect for the 20th
primes
trend
and central
ness effects, occurring with central
targets, but not with peripheral primes and central tar- percentile was not significant. This confirms the hypothgets), rather than reflecting systematic differences in re- esis that the absenceof a negative compatibility effect for
sponse activation and inhibition triggered by foveal and responsesto lateral targetswas due to the short latency of
peripheral primes. Prime-target compatibility effects many responsesand that this effect is present for slower
elicited by central and peripheral arrow primes at short responses.
Overall, the results obtained in Experiment 2 demon(0-msec)and long (128-msec)mask-target SOAs on responsesto central arrow targets were compared with ef- strate that the presenceof prime-target compatibility effects on responsesto lateral targets that were physically fects and the CPA do not critically depend on visual feadissimilar and.spatiallyseparatedfrom the primes. The tures of the target stimuli but, rather, reflect processes
results obtained in this experiment rule out the hypothe- involved in response selection and activation: Whereas
sis that the CPA is based on the different spatial separa- masked primes presentedcentrally trigger an activationtions of physically similar primes and targetsin central followed-by-inhibition process,maskedprimes presented
prime trials and peripheral prime trials: For both types of peripherally trigger an activation-only process.
targets,positive compatibility effects were obtained with
EXPERIMENT3
peripheraland centralprimes in the 0-msecSOA condition,
and a CPA was observedin the 128-msecSOA condition.
With an explanation in terms of emergent visual feaHowever, some aspectsof these results require further
consideration.First, the size of the positive compatibil- tures or perceptual repetition blindness ruled out, it is
ity effects obtained in the 0-msec SOA condition was still unclear why the spatial location of a masked prime
considerably smaller than that in Experiment lA. This should determine whether or not an inhibition process is
may have been caused by the fact that primes appeared elicited. An alternative explanation is that attentional facas targets on only 50% of all the trials. Further studies tors are responsible for the CPA. Note that in both previous experiments,targetswere presentedclose to fixation.
will have to investigate whether the strength of the initial
responseactivationdependson the probability of primes Since attention is likely to be directed to the expected lobeing presentedas targets.Second,althoughprime posi- cation of target stimuli, centrally presentedprimes were
tion X compatibility interactionswere present for both located within the attentional focus, whereas peripheral
target types in the 128-msecSOA condition, indicating primes were presentedat unattended positions. It is conthe existence of a CPA for central as well as for lateral ceivable that the presenceor absenceofresponse inhibitargets,inspection of Figure 2 suggests-and subsequent tion is directly related to the position of the prime rela/ tests confirmed-that negative compatibility effects tive to the current focus ofattention.
In the present experiment, the role of visual-spatial atelicited by central primes were significant only for central targets.This could indicatethat the physical similar- tention was studied by combining the masked prime parity between primes and targets is, in fact, crucial for the adigm with an exogenous spatial cuing paradigm. Spapresbnceofan inhibitory process,as is suggestedby the tially uninformative peripheral stimuli were presented
perceptual repetition blindne^t^t
account.There is, however, prior to each prime-mask-target sequence.These cues
an alternative explanation: As can be seenfrom Figure 2, were assumedto involuntarily summon visual-spatial atresponsesto lateral targets were considerably faster than tention to their location (Miiller & Rabbitt, 1989; Remresponsesto central targets. Mean RT to lateral targets ington, Johnston,& Yantis, 1992;Riggio & Kirsner, 1997).
was about 340 msec-that is, responsesstarted, on aver- Masked primes and targets were presented with equal
probability at cued and uncued locations. Primes preage, about 470 msec after prime offset. This primeresponse interval corresponds almost exactly to the in- sented at cued positions should be located within the
terval measured in the 64-msec SOA condition in Ex- focus of attention, whereas primes presented at uncued
periment lA, where mean RT to central targetswas about positionsshouldbe locatedoutsidethis focus. If the CPA
400 msec and no prime-target compatibility effects were dependedon the current locus oftransient visual-spatial
attention, in that response inhibition was elicited for
obtained at all (seeFigure l, left). Afacilitation-followedby-inhibition account would assume that the fastest re- primes at attended locations but not for primes at unat-
compatibility interactionsat the 128-msecSOA, indicating the existence of a CPA, were significant for central
as well as for lateral targets [RT: Fs(I,7) : 69.22 and
20.29,ps < .001 and .003, for central and lateral targets,
respectively;error rates:.Fs(1,7): 23.02 and7.29,ps <
. 0 0 2a n d . 0 3 l l .
1374
SCHLAGHECKENAND EIMER
tended locations, a negative compatibility effect similar
to the effects obtained previously with central primes
should be found for trials in which peripheral masked
primes appeared at cued locations, wheieas a positive
compatibility effect should be obtained at uncued (unattended) locations.
Note that a failure to find this pattern of results could
be due to the fact that the peripheral cues were not effective in attracting visual-spatial attention. Thus, in order
to have an independent estimate of the effectiveness of
the exogenous attentional cuing procedure, occasional
probe trials were delivered together with the usual prime
trials. In probe trials, participants had to respona tb tettpointing or right-pointing arrows, which appeared instead of the prime but were neither masked nor followed
by an additional arrow. Ifthe attentional cuing procedure
was effective, responses to probe targets at cued locations should be faster than responses to probe targets at
uncued locations. If such exogenous cuing effects were
obtained on probe trials without any modulatory influence
of attentional cuing on prime-target compatibility effects, this would demonstrate that visual-spatial attention is not a critical factor for the presence of the CpA.
In the practiceblock aswell as in the baselineblock, prime trials
and probe trials were presentedwithout a prior cuing stimulus. In
the attentionalcuing block, eachtrial startedwith the presentation
ofa peripheralcue,appearingrandomlyandwith equaiprobability
either 3.2'above or 3.2obelow fixation and remaining visible for
I l0 msec. Either a prime trial or a probe trial (seeabove)was delivered 55 msec after the offset of the cue. Masked primes and targets appearedwith equalprobability at cued and uncuedlocations.
Data analysis. For prime trials in the baselineblock, two-tailed
I testswere computedon RTs and error ratesfor compatibleversus
incompatibletrials. For prime trials in the attentionalcuing block,
repeatedmeasuresANOVAs wereperformedon RTs and error rates
for the within-subjectsfactorsofcue validity (cued vs. uncuedlocation), prime/targetcompatibility (compatible or incompatible),
and responseside (left or right) For probe trials in the attentional
cuing block, two-tailedttests werecomputedon RTs anderror rates
for targetsat cuedversusuncuedlocations.
Results
Cue validity effects. Probe trials in the attentional
cuing block showed a main effect of cue validity on RT
and error rates, indicating that the cues were efficient in
attracting visual-spatial attention. RTs were 441 and
471 msecto targetsat cued and uncuedlocations lt(j) :
3 .43,p < .01I ], and the respectiveerror rates were | .3Vo
and3.9o/ott(1): 4.24,p<.0041.Cue validiryalsoaffected
Method
RT (but not error rate) on prime trials, since responseson
Participants. Eight paid volunteers(4 male), 23-39 yearsofage
validly cued prime trials were about 12 msec faster than
.
(meanage,26.6 years),participatedin the experiment.Al1 the par- responses
on invalidly cued prime trials [F'(I,7) : 6.9j ,
ticipantswereright-handedand had normal or corrected-to-normal
p < .033; seeFigure 31.
vision.
Prime-target compatibility effects. In the baseline
Stimuli and Apparatus. Left- and right-pointing doublearrows
block, in which no cues were presented,performance
(<< and >>), subtendinga visual angleofapproximately1.15.x
0.4o,servedas prime and targetstimuli. The mask was constructed was better on compatibletrials than on incompatibletrivia superimposingthese stimuli upon one another.The cue con- als [RT: 424 msecvs. 445 msec; t(7) : 3.91,p < .006;
sistedof a rectangularframe of approximately1.4. X 0.8.. In addi- error rates:5.3% vs. 92%:
t(7) : 3 .15,p < .0161.Analtion, a fixation crosswaspresented,subtendinga visual angleofapogous effects were found in the attentional cuing block
proximately0.3oX 0.3o.All the stimuli werepresented
in blackon (see
Figure 3), since compatible trials produced faster
a white backgroundon a | 7-in. computerscreen.
RTs
and
fewer errors than did incompatible trials [RT:
Procedure. The generalexperimentalset-upwas similar to ExperimentsI and2. To supportcentral eye fixation, a fixation cross F(|,7) : 35.87,p < .00I ; error rate:F(|,7) : 30.33,p <
was continuouslypresentat the center of the screen.The experi- .001]. Moreover,the positive compatibility effect observed
ment startedwith a practiceblock of 160 trials. The data from this in the attentionalcuing block showed a tendency
to be
block werenot stored.After the practiceblock, a baselineblock was larger
on valid trials than on invalid trials, confirmed by
delivered(240 trials), in which no cueswere presented.Finally, an
a validity x compatibility interaction that was significant
attentional cuingblock was delivered (960 trials). In each block.
therewasa breakafterevery80 trials.The participantsinitiatedthe for error rate lF(|,7) : 32.42,p < .001I and approached
significancefor RT lF(|,7) : 4.63,p < .0691.
next seriesof 80 trials when they felt ready to do so.
In all the blocks, 80% of all the trials were prime trials and20Vo
of all the trials wereprobetrials.Primetrials consistedof a prime
arrow (16-msecduration),pointing randomly and with equalprobability to the right or to the left. The prime was immediately followed by a mask (100-msecduration), which in turn was immediately followed by a targetstimulus(100-msecduration).Trials were
terrned,compatibleif prime and target arrows pointed in the same
direction and incompatibleotherwise.Both conditionswere equiprobableand randomizedwithin eachblock. probe trials consisied
ofone doublearrow only (probearrow), which was neithermasked
nor followedby an additionalarrow andwaspresentedfor 100 msec.
The intertrial interval was I sec.The participantshad to respondto
the probe arrows in the sameway that they had to respondto the
prime trial targets.For any given trial, all the stimuli (prime, mask,
and target in the prime trials and probe arrows in the probe trials)
appearedeither3.2oaboveor 3.2obelow fixation, both positions
being equiprobableand randomizedwithin eachblock.
Discussion of Experiment 3
If the responsetendenciestriggered by stimuli within
the attentional focus are subject to inhibition, whereas
no inhibition is elicited when suchtendenciesare caused
by unattended stimuli, presenting masked primes and
targetsat previously cued or uncuedperipherallocations
should affect the direction of prime-target compatibility
effects. In Experiment 3, uninformative peripheral cues
were presentedeither aboveor below fixation, and all subsequentstimuli (prime, mask, and target on prime trials
and probe arrows on probe trials) appearedrandomly and
with equal probability at the cued location or at the uncued location. If the CPA was determinedby the current
focus of attention, a negative compatibility effect similar
CENTRAL-PERIPHERALASYMMETRYIN MASKEDPRIMING
1375
RT (ms)
450
440
I
u
430
420
410
400
390
380
15
o
370
1 0 q
360
5
350
0
compatible
compatible
d
d
3
incompatible
Figure 3. Experiment 3: Performance in compatible and incompatible prime trials in the
experimental block. Lines depict reaction times (RTs), bars depict error rates. Invalidly cued
trials are indicated by white circles and white bars; validly cued trials are indicated by black
squares and black bars.
to the effects observedwith centralprimes should be found
for maskedprimes and targetsat cued locations,whereas
a positive compatibilityeffect should be presentat uncued locations.
The results of the presentexperiment do nor supporr
this assumption.Although exogenouscuing effectson
RT and error rates on probe trials showedthat attention
was summonedto cued locations,therewas no indication
that a negativecompatibility effect was elicited for prime
trials at these locations.On the contrary,the interaction
betweencue validity and compatibility (only marginally
significant for R! but highly significant for error rate)
reflected the opposite direction of effects; If anything,
the positive compatibility effect was largeron valid triais
than on invalid trials.6Thus, one is led to assumethat attentional factors play no major role for the occurrenceof
the CPA.
Howeveq it has to be notedthat in the Dresenttrial-bvtrial cuing experiment,only the roleof transient attentioial
shifts was investigated.In the previousexperiments,targets usually appearedat fixation, thus presumably resulting in a sustainedfocusing ofattention at the screen
center.Since there is some evidencethat effects oftransient attention are different from effectsofsustained attention (e.g.,Eimer, 1996;Nakayama& Mackeben,1989;
Posner,Snyder,& Davidson, 1980),the possibility remains
that the location of prime stimuli, relative to a sustained
focus of spatial attention,plays a role for the CpA. Alternatively,the CPA might be directly relatedto the retinal eccentricity of the maskedprimes. In Experiment 4,
theseissueswere investisated.
EXPERIMENT
4
In this experiment,the influence of sustainedvisualspatial attention,as well as the relationshipbetweenthe
CPA and the retinal eccentricity of the masked primes,
was investigated.Since the CPA was found to be unaffected by transient attentional orienting in the previous
experiment,it was testedwhethersustainedattentionhas
an effect on the CPA. To investigatethis iSsue,the possibility of maintaininga narrow focus of attentionin order
to selectivelyfilter out peripherally presentedirrelevant
information was manipulatedby varying the conditions
of prime presentation.In previous experimentsemploying
peripheralprimes, prime location was constantthroughout an experimentalblock. It should be more difficult to
maintain a narrow focus of sustainedattention when irrelevantstimuli are presentedat unpredictablelocations
than when their location is constantand, therefore, en-
1316
SCHLAGHECKENAND EIMER
tirely predictable(seeAllport, 1989,for the idea that the
attentional filtering of irrelevant stimuli can depend on
the availability and reliability ofadditional spatialinformation about thesestimuli).
The participantswere assignedto one ofthree groups.
For Group A, primes were presented either centrally or
unilaterally displaced from fixation, with each of l3 possible prime locations (see below) randomized within
eachblock. This group was expectedto have the greatest
difficulties in maintaining a narrow focus of attention.
For Group B, masked primes were presentedbilaterally,
thus reducing the number of possibleprime locations in
any given block to 7. It was assumedthat becauseof the
greater uniformity of stimulus presentation, maintaining
a narrow attentional focus would be less disrupted in this
condition. For Group C, primes were presentedbilaterally
and prime eccentricity was held constant throughout a
given block. The participants in this group were expected
to have the least difficulty selectively filtering out irrelevantperipheralinformation.
If this between-subjectsmanipulation has the predicted effect on sustainedattention, the participants in
Group A (most diffuse spatial attention) should show the
longest RTs in responseto targets,whereasthe participants in Group C (most focused spatial attention) should
show the shortestRTs. If the CPA is sensitiveto the possibility of selectively filtering out peripheral masked
primes,Group A should show the smallestCPA, because
with diffuse spatial attention, the difference between attended (central) and unattended (peripheral) prime locations should be relatively small. Group C, on the other
hand, should show the largestCPA, becausewith highly
focused attention, the attentional difference between
central and peripheral locations should be most pronounced.In contrast, ifthe CPA is not affected by sustained attention, no CPA differences should be found between groups.
Rather than being due to attentional factors, the CPA
may primarily reflect physiological characteristicsof the
visual system.Since the perceptualsensitivity of the visual system declines with retinal eccentricity (Anstis,
1974;DeValois& DeValois, 1988;Lie, 1980;Rijsdijk,
Kroon, & van der Wildt, 1980),a stimulus presentedperipherally has a different-probably lower-representational quality than does a stimulus presentedat fixation.
It seems possible that the strength of perceptual representationsof prime stimuli is reflected in the strength of
the primed responseactivation and that only motor activations exceedinga certain thresholdare subjectto inhibition. If this were the case,it would be more likely that
stimuli presentedwithin regions of high perceptualsensitivity would eventually give rise to an inhibition process.
Negative compatibility effects may thus occur with central primes becausethey are presented foveally, whereas
positive compatibility effects may occur for peripheral
primes becausethey are presented extrafoveally.
Following this line of argument,one may expectto find
direct relationships between variations of the retinal position of the primes and the magnitude and direction of
priming effects:With increasingretinal eccentricity ofthe
prime, a decreasingnegativecompatibility effect should
be obtained, which should turn into a positive compatibility effect beyond some critical distancefrom the fovea.
Sincethe perceptualsensitivity ofthe retina declinesfaster
along the vertical axis than along the horizontal axis
(i.e., the area of maximal sensitivity has an elliptical
shape,with the horizontal meridian determining its major
axis; see,e.g.,Rijsdijk et al., 1980),this critical distance
should be reachedat smaller eccentricities with vertically
displacedprimes than with horizontally displacedprimes.
To test this issue,maskedprimes either were presented
at fixation or were horizontally or vertically displaced by
0 . 5 5 ' , 1 . 1 " ,1 . 6 5 " , 2 . 2 " , 2 . 7 5 o , o3r. 3 oo f v i s u a la n g l e .
Note that in contrast to the hypothesis that the CPA depends on physiological inhomogeneities of the retina, the
idea that the CPA is due to attentional factors would not
predict any difference betweenhorizontally and vertically
displacedprimes in the decreaseand reversalof priming
effects with increasing retinal eccentricity. Although attentional selectivity generally increaseswith increasing
retinal eccentricity of irrelevant information, this effect
is generally found to be equivalent for horizontally and
vertically displaced distractors, and the attentional field
therefore is assumedto be circular in shape(e.g., Henderson& MacQuistan,1993;Hughes&Zimba,1987; Pan
& Eriksen,1993).
Method
(13 male),20-32
paidvolunteers
Participants.Twenty-four
in theexperiment.
yearsofage(meanage,23.7years),participated
vision,and
hadnormalor corrected-to-normal
All theparticipants
to
wereassigned
The participants
all but onewereright-handed.
oneof threegroups(GroupA, GroupB, andGroupC) in orderof
in eachgroup.
Therewere8 participants
theirappearance.
wereidentiStimuliand Apparatus.Thestimuliandapparatus
ofa fixation
3, exceptfor the absence
cal to thosein Experiment
crossanda cue.
set-upwasidenticalto those
Procedure.Thegeneralexperimental
I to 3, exceptthat the targetstimuliwerealways
in Experiments
at
presented
at fixation,whereasmaskedprimeseitherappeared
by 0.55',1.1",1.65',2.2",2.75', or 3.3"
fixationor weredisplaced
waseitherhorizontalor vertical,and
ofvisualangle.Displacement
wasalwaysblocked.
thedirectionof displacement
uniprimesandmaskswerepresented
ForGroupA, peripheral
blocks,theycouldappearatfixationandalong
laterally.In separate
the verticalmeridianaboveor belowfixation or at fixation and
alongthe horizontalmeridianto the left or right of fixation.For
biprimesandmaskswerepresented
GroupsB andC, peripheral
laterally-thatis, eitheraboveandbelowfixationor to theleft and
from fixation.Moreover,for
right of fixation,andequidistant
(0'-3.3") wererandomized
GroupsA andB, primeeccentricities
in bothgroupsreceived
within eachblock.Halfofthe participants
five horizontalblocksfollowedby five verticalblocks(eachblock
was
for theotherhalf,thissequence
336trials),whereas
containing
wereblocked,andthe
ForGroupC, primeeccentricities
reversed.
240trials)wasrandomized
sequence
of l4 blocks(eachcontaining
CENTRAL-PERIPHERALASYMMETRYIN MASKED PRIMING
1377
Primes horizontally displaced
410
(tt
5
F
TT
390
370
o
2 0 d
ln
ds
2.2 2.75
1.65. 2.2" 2.75. 3.3.
0
Primesverticallydisplaced
410
aon
o
-E
-(r
370
o
2 0 d
r v s
o"
0.55" 1.1' 1.6s'2.2.2.75
3.3
0.
0.55" 1.1. 1.65"2.2 2.75" 3.3.
O" 0.5s" 1.1" 1.65"2.2 2.75" 3.3.
0
Figure 4. Experiment 4: Reaction times (RTs; lines) and error rates (bars) for compatible trials (gray) and incompatible trials
(white), plotted separatelyfor the three experimental groups. Upper panel: Blocks with horizontally displacedprimes. Lower panel:
Blocks with vertically displacedprimes.
for eachparticipant.Therewasa total of 120trials for eachofthe
28 conditions(direction ofdisplacementX eccentricity x compatibility) throughout the experiment.A rest period was included in
the middle of each block, and the participantsinitiated the second
halfofthe block whenthey felt readyto do so.
Data analysis. RepeatedmeasuresANOVAs wereperformedon
RTs and error rates for the between-subjectsfactor of group
(Groups A, B, and C) and for the within-subjectsfactors of prime
position(0',0.55", 1.1",1.65',2.2',2.75",and 3.3'),directionof
displacement(horizontalor vertical),andcompatibility (compatible
or incompatible).Where appropriate,Greenhouse-Geisser
adjustments to the degreesof freedomwere performed (indicated in the
resultsectionby e).
e: .549, respectively].Second,RTs were faster and
fewer errors were made with vertically displaced primes
than with horizontally displaced primes IF(|,21) :
10.35, p < .004, andF(1,21) : 4.47,p < .0471.Theseeffects interacted for both RTs and error rates lF(6,126) :
2 . 9 6 ,p < . 0 3 2 ,e : . 5 7 1 ,a n dF ( 6 , 1 2 6 ): 5 . 0 0 , p < . 0 0 1 ,
e : .679, respectively], since the performance increment
with increasing eccentricities of the masked primes was
more pronounced for vertically displaced primes than
for horizontally displacedprimes. Interactionsof these
effects with the factor of group are reported below.
Prime-target compatibility. Generally, RTs were
faster and fewer errors were made on incompatible than
Results
on compatible
t r i a l s[ R T : F ( 1 , 2 1 ) : l l l . 9 5 , p < . 0 0 1 ;
Prime position. Figure 4 shows RTs and error rates errorrate: F(\,21) : 61.25,p <.001, respectively].Howobtained in compatible and incompatible trials for dif- ever, theseeffects becamesmaller with increasing retinal
ferent prime eccentricitiesand directions of displacement, eccentricitiesof the masked primes [RT: .F(6,126):
displayed separately for the three experimental groups. 88.95,p < .001,e : .390;error rate:F(6,126) : 44.45,
RTs became generally faster and fewer errors were made p <.001, e : .383]. Furthermore,the negativecompatiwith increasing eccentricity of the prime [F(6,126) :
bility effect was generally smaller for vertically displaced
24.47,p < .00l, e : .565,andF(6,126): 33.83,p < .00l, primes than for horizontally displaced primes [RT:
I378
SCHLAGHECKENAND EIMER
F(\,21) : 27.28,p < .001; error rate:F(I,21) : 12.96, tion, Group C profited from increasingthe distancebep < .002l.Inspection of Figure 4 revealsthat this inter- tween the target and the peripheral masked primes,
preaction was due to the fact that with horizontally displaced whereas Group A did not, which also indicates that
to
selecability
the
affected
prime
location
of
primes, the negativecompatibility effect decreasedslowly dictability
Howand vanishedat the largest eccentricity, whereaswith ver- tively filter out irrelevant peripheral information.
groups with
these
between
no
difference
was
there
ever,
and
rapidly
more
primes,
it
decreased
tically displaced
indicates
turned into a positive compatibility effect at2.2" [three- respectto the pattern of priming effects. This
efattentional
by
differential
not
caused
is
way compatibility x eccentricity X direction of displace- thai the CPA
primes
peripheral
and
ofcentral
processing
<
.
0
0
1
,
the
fects
on
e
:
; T: F(6,126): l0.3l,p
m e n t i n t e r a c t i o nR
a nar.542',error rates:F(6,126) : 4.52,p < .004, e : .5701. under conditions in which it is possible to maintain
exAdditional / tests,conductedfor each prime position row central focus ofvisual attention throughout an
separately,revealedthe following pattern of effects. Hor- perimentalblock.
The main result of the present experiment, however, is
izontally displaced primes causeda significant negative
finding of a strong dependency between the magnithe
between
retinal
eccentricities
compatibility effect at all
<
tude
and direction of the priming effect and the retinal
n
o
>
A
t
3
.
3
"
,
.
0
1
2
]
.
a
l
l
p
s
2
.
7
5
,
r
s
(
2
3
)
2
.
7
5
o
0'and
[all
primes, the
priming effect was present.Vertically displacedprimes position of the maskedprimes: With central
since
observed,
was
effect
compatibility
negative
usual
causeda significant negative compatibility effect at retiincomon
made
were
fewer
errors
>
and
faster
RTs
were
6.01'
/s(23)
1.1"
between0" and
nal eccentricities
[all
was
all ps < .0011.At 1.65', only a nonsignificant negative patible trials than on compatible trials. This effect
eccentriciretinal
:
increasing
with
| -'/5,p < .0941. continuously reduced
compatibility effect was obtained [t(23)
on
At retinal eccentricitiesbetween2.2o and3.3", vertically ties. Moreover, the slope of this reduction depended
that
5
shows
Figure
prime
displacement'
of
the
direction
positive
compatidisplacedprimes causeda significant
with horizontally displacedprimes, negativecompatibilbility effect [all rs(23)> 2;77, allps < .012].
and
Sustained attention. None of the effects that included ity effects on RTs and error rates decreasedslowly
(3.3'
visual
of
eccentricity
largest
the
at
only
compatibility as a factor showed any interaction with the vinished
of the
factor of group (all Fs < 2.2). This is especiallyimpor- angle). With vertically displacedprimes, the slopes
retinal
a
zero
at
reaching
steeper,
were
functions
reduction
group
on
of
effect
main
a
significant
tant since there was
increasRT [F(2,21) : 4.18,p < .030],with Group A producing eccentricity of approximately2o and becoming
efcompatibility
negative
(i.e.,
initial
the
negative
the longest RTs and Group C the shortest (mean RTs ingly
larger
with
effect)
positive
compatibility
into
a
turned
feit
reC,
A,
B,
and
were 375, 358, and 345 msec for Groups
spectively),indicating that the manipulation of prime lo- eccentricities.
The fact that the critical eccentricity where the negacition predictability successfully influenced attentional
tive compatibility effect turned into a positive compatifocusing.
bility effect was reachedearlier with vertically displaced
of
disMoreover,effects of eccentricity and direction
primes than with horizontally displaced primes can be
increperformance
group:
The
placementinteractedwith
is priment with increasingretinal eccentricity of the masked iaken as strong evidence for the idea that the CPA
the
visual
of
properties
physiological
primes was largest for Group C, smaller for Group B' marily based on
the
and absentfor Group A [RT: F(12j26) : 5-98,p < .001, system rather than on attentional factors. Although
along
faster
changes
retina
ofthe
<
sensitivity
perceptual
p
e
:
.
0
0
2
,
3
.
8
9
,
e : . 5 6 5 ;e r r o r r a t e s :F ( 1 2 , 1 2 6 ) :
(Rijsdijk
.549]. Second,there was a group X direction of dis- ihe vertical axis than along the horizontal axis
depensuch
no
shows
:
selectivity
<
et al., I 980), attentional
12.08,P
placement interaction on RTs lF(2,21)
&Zimba,
Hughes
1993;
MacQuistan,
(Henderson
&
dency
.001], since the RT advantagefor vertically displaced
primes,as comparedwith horizontally displacedprimes' 1987;Pan & Eriksen,1993).
was present only for Group B and Group C, but not for
GENERAL DISCUSSION
Group A. Finally, there was a three-way interaction of
group x eccentricity X direction of displacementinterThe nature and role of inhibitory mechanisms in the
i c t i o n o n R T s [ F ( 1 2 , 1 2 6 ): 2 . 2 0 ,p < . 0 4 6 ,e : . 5 7 1 ] .
adaptivecontrol ofcognitive processesand behavior have
been studied intensively in a variety of paradigms, leadDiscussion of Experiment 4
The present results provide additional evidence that ing to the concept of inhibitory control of information
this concepthas
the CPA does not depend on attentional factors. The ob- processing.Usually, evidencesupporting
stimuli
suprathreshold
employing
in
studies
found
groups
Leen
the
three
between
servedperformance differences
et al',
Raymond
1985;
Maylor,
showed that the manipulation of sustainedattention was (e.g., Kanwisher, 1987;
near
presented
are
stimuli
When
l9S5).
1992;Tipper,
successful:Group A (completely randomized prime preprocesses,
facilitatory
for
evidence
threshold,
least
detection
the
(indicating
sentation)producedthe longestRTs
generally been found
focused spatial attention), whereas Group C (completely but not for inhibitory processes,has
McCormick,
blocked prime presentation) produced the shortest RTs (e.g., Allport et al., 1985; Marcel, 1980;
in
of
experiments,
series
in
a
recent
However,
(indicating strongly focused spatial attention). In addi- lgtT.
CENTRAL-PERIPHERALASYMMETRYIN MASKED PRIMING
PrimeDisplacement
*
1379
15
horizontal +
o
o
o-
10#
o
f
o
o
c)
o
3
t r E
s g )
d.
o
o
I
^V
0
='
o
o
3
E
g)
=.
o
o
-10
s
-20
-5
0
0.55
1.1 1.65 2.2
Eccentricity
2.75
3.3
Figure 5. Experiment 4: Mean compatible-incompatible
differences in response time
(RT; thick lines with square marks) and error rates (thin lines with circular marks), plotted separately for vertically displaced primes (white) and horizontally displaced primes
(black). RT differences and error differences are collapsed across groups.
which target stimuli were precededby centrally presentedmasked primes, evidencefor the existenceof inhibitory control processesunder near-thresholdconditions has been reported. Positive compatibility effects
(betterperformancein compatibletrials, in which primes
and targetswere identical,than in incompatibletrials, in
which they were mappedto oppositeresponses),occurring when the interval betweenprime and target presentation was short, were found to turn into nesative compatibility effects (better performancein in-compatible
trials) when this interval was longer. It was argued that
this pattern of resultsreflectsan initial facilitation of the
responseassignedto the prime that is subsequentlyinhibited. The puzzling fact that no evidencefor response
inhibition was found when the maskedprimes were presentedperipherally (CPA) suggestedqualitativelydifferent types of perceptuomotorprocessingfor foveal and
peripheral stimuli. The aim of the present seriesof experimentswas to investigatethis idea and to test alternative accountsof the CPA.
Experiment I tested three explanationsfor the presenceof a CPA (delayedactivation,delayedinhibition, or
absentinhibition ofresponsetendenciestriggeredby peripheral primes) by varying mask-targetSOA between0
and 192msec for centraland for peripheralprimes. With
short prime-target SOAs (0 and 32 msec),positive compatibility effects were obtained for central and for peripheral primes. This is inconsistentwith the idea that the CpA
is causedby a delayedactivationofthe responsetendency
triggeredby peripheralprimes.With long SOAs(96 msec
and longer), negativecompatibility effects were observed
for centralprimes,whereaspositivecompatibility effects
remained present for peripheral primes. This CpA remained stablewithin the SOA range investigatedin Experiment l. The absenceof a negativecompatibility effect for peripheral primes at long SOAs contradictsthe
hypothesisthat responseinhibition is simply delayedfor
peripheralprimes.In terms of responseactivationand inhibition, this pattern ofresults suggeststhat both central
and peripheralprimes initially trigger a partial response
activation,presumablymediatedby direct perceptuomotor
links. With central primes, this initial activation is subsequentlyinhibited,andthis inhibitory phasedoesnot dissipaterapidly but remainspresentfor at least 100 msec.
No such inhibition is elicited for peripheral primes, for
which the initial responsetendencytriggered by the prime
affectsperformancefor at least200 msec.In other words,
activation-followed-by-inhibition
is restrictedto centrally
I38O
SCHLAGHECKENAND EIMER
presentedprimes, whereasactivation-only is elicited by
peripheralprimes.
Experiment 2 ruled out the idea that these effects are
due primarily to perceptual properties of the stimuli (resulting in specific emergent features and/or repetition
blindnesseffects),rather than reflecting responsefacilitation and inhibition: Positive compatibility effects for
central and peripheral primes with a O-msecmask*target
SOA, as well as a CPA with a 128-msecSOA. were obtained not only for responsesto central targetsthat were
physically similar to the prime stimuli, but also for responsesto lateral targetsthat were physically dissimilar
from the masked arrow primes. Note that the fact that
prime-target compatibility effects were elicited independentof the visual similarity betweenprimes and targetsdistinguishes them from other inhibitory effects, such as
repetition blindness and the attentional blink, where such
similarities apparently play an important role (see Kanwisher & Potter, 1990; Raymond, Shapiro, & Arnell,
1995).
Together with LRP data obtained in previous studies
(Eimer, 1999; Eimer & Schlaghecken,1998), these experimentsprovide convergingevidencethat briefly presentedand subsequentlymasked stimuli trigger apartial
responseactivation and that this initial response tendency is later reversedin the caseofcentrally presented
primes. Second,and most important, the CPA cannotbe
explained simply by differencesin the timing of effects
ofcentral and peripheralstimuli on responsestages(delayed activation or delayed inhibition) but seemsto reflect qualitative differences in perceptuomotor links, with
inhibition restrictedto responseactivationby foveal events.
The results from Experiments I and 2 thus suggestthat
with respectto responseactivationand inhibition, foveal
eventsare "special," but they do not provide information
about what makes them special. Experiments 3 and 4 investigatedthe role of visual-spatial attention for the CpA:
Responseinhibition may have been restricted to foveal
primes becauseattentionwas focusedat fixation and responseinhibition processesare elicited only by stimuli
presentedinside the focus of visual-spatialattention.We
tested this by manipulating the focus of transient (Experiment 3) and sustained(Experiment 4) attention.
Converging evidencewas found that the CPA is not modulated by attentional factors. Neither transient shifts of
focal attention triggered by exogenous trial-by-trial cuing nor differencesin the focus ofsustained attentionrelated to the predictability of prime locations had an effect on the direction and size of the CPA.
Alternatively, responseinhibition may be restricted to
foveal primes becauseresponseinhibition processesare
elicited only by stimuli presentedat retinal locations of
high perceptualsensitivity.Experiment 4 revealeda strong
dependencybetween the perceptual sensitivity ofthe retinal region ofprime presentationand the size and direction
of priming effects. Generally, there was a gradual decreasein the size of negativecompatibility effects with
increasing retinal eccentricity of the prime, parallel to
the gradual decreaseofperceptual sensitivity at more peripheral retinal positions (Anstis, 1974:DeYalois & DeValois, 1988; Lie, 1980).Most important, this decrease
was steeperwith vertically displaced primes than with
horizontally displacedprimes, parallel to the faster decreaseof perceptualsensitivity along the vertical meridian
than along the horizontalmeridian (Rijsdijk et al., 1980).
This pattern of results suggeststhat the CPA is closely
linked to variationsin perceptualsensitivity: An inhibition of a responsetendencytriggeredby a maskedprime
is elicited only if the prime stimulus is presentedat retinal locationsofhigh perceptualsensitivity,regardlessof
whether or not this location is attended.
Howeveq these findings still leave open the question
of why the retinal location of a masked prime should determine the presenceor absenceof responseinhibition.
One tentative explanation is that responseinhibition processeswill be initiated only when the primed response
has reachedan activationthreshold: Given the existence
of direct perceptuomotor links, strong perceptual representationsmay elicit strongcr motor activations,which
will be inhibited if they reach above-thresholdvalues,
whereas weak perceptual representationswill elicit
weaker motor activations, which remain below inhibition
threshold. Presumably,the representationalquality of
sensorytraces elicited by prime stimuli decreaseswith
increasing retinal eccentricity. Consequently,motor tendencies triggered by foveal primes are more likely to be
actively inhibited than motor tendenciestriggered by peripheral primes. Similar ideas of a threshold determining
whether gradual-accumulation-of-activationor activationfollowed-by-inhibition will take place were proposed by
Hagenziekerand van der Heijden (1990; seealso Hagenzieker, van der Heijden, & Hagenaar,1990).
Several questions are raised by this tentative idea. For
example,although a strong link betweenCPA and physiological properties of the visual systemhasbeen demonstrated, it is unclear whether the relevant property really
is perceptual sensitivity, rather than some other factor
that varies with retinal eccentricity, such as, for example,
temporal integration. Moreover, the threshold idea assumesthat there are no qualitative differences in the processing of central and peripheral information-that is,
that in either case, the processing is performed by the
samevisuomotor pathways.This view is not necessarily
correct. For example, there is some evidence that the
amount of retinal ganglion cells projecting to the superior
colliculus increaseswith increasingretinal eccentricity
(Bunt, Hendrickson,Lund, Lund, & Fuchs, 1975; perry
& Cowey, 1984). One could therefore assumethat different visual processing streamscontribute differentially
to the processing of central and peripheral stimuli.
Whereasvisuomotor systemsoperative ingaze shift control (e.g.,the superiorcolliculus and relatedmotor structures) may be more involved in processingperipheral
stimuli than in processing central stimuli, even under
conditions in which no overt eye movements are performed, perceptuomotorsystemsresponsiblefor visuo-
CENTRAL-PERIPHERALASYMMETRYIN MASKEDPRIMING
manual control may receive their input primarily from
the fovea. The CPA may be related to such functional differencesbetweenvisuomotor subsystems.
However, at least two predictions can be derived from
the responseactivationthresholdconceptthat can be tested
in further studies.Ifthe direction and time courseofpriming effects on performancedependon the degreeto which
the primed responseis activated,centrallypresentedprimes
that elicit only a weak sensorytrace (e.g., becauseof a low
signal-to-noiseratio) should give rise to positive compatibility effects, reflecting the absenceofresponse inhibition. Similarly, peripheralprimes that elicit a sufficiently
strong activation (e.g., owing to size scaling or longer
presentation times) may trigger activation-followed-byinhibition and, thus, a negativecompatibility effect.
I38I
GooLxrsrnN, P. (1997). Size scaling and spatial factors in visual attention. A merican Journa l of Psycho l oglt, 110, 397-4 15.
GurroN, G., Colrs, M. G. H., Snsvllc, E. J., Enrxsnx, C. W., &
DoNcuIN, E. (1988). Pre- and post-stimulus activation ofresponse
channefs: A psychophysiological analysis. Journal of Experimental
Psychology: Human Perception & Performance, 14,331-344.
HecnNzrnxrn, M. P., & vAN DERHEUDEN,A. H. C. (1990). Timecourses in visual-information processing: Some theoretical considerations. Psychological Research, 52, 5-12.
HlcnNzrErnn, M. P., vnN onn HruneN, A. H. C., & HlcpNaen, R.
(1990). Time-courses in visual-information processing: Some empirical evidence for inhibition. Psychological Research, 52, 13-21.
HrNnEnsoN, M. J., & MrcQursrrN, A. D. (1993). The spatial distribution ofattention following an exogenouscue. Perception & Psychophysics,53,22l-230.
Hocnnlus, L., & JosNsroN, J. C. (1996). Perceptual repetition blindness effects. Journal ofExperimental Psychology: Human Perception & Performance, 22, 3 55-366.
Houcuron, G., & Trrern, S. P. (1996). Inhibitory mechanisms ofneural
and cognitive control: Applications to selective attention and sequenREFERENCES
tial action. Brain & Cognition,30,20-43.
HucHEs, H. C., & Ztwnt,L. D. (1987). Natural boundaries for the
ALrponr, A. (1989). yisual attention.In M. I. Posner(Ed.), Foundaspread of directed visual attention. Neuropsychologia, 2, 5-18.
tions of cognitive science(pp.63l-682). Cambridge, MA: MIT Press. KlNwtsHrn, N. (1987). Repetition blindness: Type recognition without
perceptual
Allponr, A., Trppnn,S., & CHrrlrpl, N. R. J. (1985).
intetoken identification. Cognition,2T , 117-143.
gration and postcategoricalfiltering. In M. I. Posner & O. S. M.
KeNwrsnnn, N., & PorrEn, M. C. (1990). Repetitionblindness:Levels
Marin (Eds.),lltention and performance XI (pp. 107-132). Hillsdale,
of processing."/ournal of Experimental Psychology: Human PercepNJ: Erlbaum.
tion & Performance,16,30-47 .
(1974).
ANsrrs, S. M.
A chart demonstrating variations in acuity with
Klorz, W., & Wor-n4 P. (1995). The etrect of a masked stimulus on the
retinal position. Vision Research, 14, 589-592.
response to the masking stimulus. Psyclrological Research, 58, 92- I 0 I .
AnnurHNorr, K. D. (1995). Inhibitory mechanismsin cognition: pheKoee, B., Rrsr, F., & MrrrrEn, U. (1996). N200 in the flanker task as
nomena aiid models. Current Psychology of Cognition, 14, 3-45.
a neurobehavioral tool for investigating executive control. PsychoBurr, A. H., HsNonrcxsoN,A.8., LuNo, J. S., I-urvo,R. D., & FucHs,
p hysio logy, 33, 282-294.
A. F. (1975). Monkey retinal ganglion cells: Morphometric analysis LIr, I. (1980). Visual detection and resolution as a function
ofretinal
and tracing ofaxonal projections, with a consideration ofthe peroxlocus. Vision Research, 20, 967 -974.
idase technique. Journal of Comparative Neurology, 164, 265-285.
MlncEI-, T. ( I 980). Conscious and preconscious recognition ofpolyseCoyln, K. (1994). The relationshipbetweentime ofarrival ofnontarmous words: Locating the selective effects ofprior verbal context. In
gets and their spatial location: Evidence for asymmetries in visual atR. S. Nickerson (Ed.),Attention and performance ITIII Qtp.415-457).
tentional processing. Perception, 23, 849-856.
Hillsdale, NJ: Erlbaum.
D n c e N n a c H , D . , C n n n , T . H . , & W r l u E r - t r s n N ,A . L . ( 1 9 8 9 ) . T a s k - Mlv, C. P., KrNr, M. J., & HesHnn, L. (1995). Determinants
of negainduced strategiesand near-thresholdpriming: Conscious influences
tive priming. Psychological Bulletin, ll8, 35-54.
on unconscious perception.Journal of Memory & Language,28,
Mlylon, E. A. (1985). Facilitatory and inhibitory components of ori412-443.
enting in visual space.In M. I. Posner& O. S. M. Marin (Eds.),lrDB JoNc, R., Cor-Es,M., & LoclN, G. D. (1995). Strategiesand mechtention and performance XI (pp. 189-204). Hillsdale, NJ: Erlbaum.
anisms in nonselective and selective inhibitory motor control../oarMcConurcr, P.A. (1997). Orienting attention without awareness../orrnal ofExperimental Psychologlt: Human Perception & Performance,
nal ofExperimental Psychology: Human Perception & performance,
2 1 , 4 9 8 - 5I l .
23,168-180.
DsVrll-ors, R. L., & DEVIr-ors, K. K. ( 1988). Spatial vision. New york:
MII-lEn, J. (1991). The flanker compatibility effect as a function of visual
Oxford University Press.
angle, attentional focus, visual hansients, and perceptual load: A search
Eturn, M. (1996). ERP modulationsindicatethe selectiveprocessinsof
for boundary conditions. Perception & P sychophysics, 49, 2'l 0-288.
visual stimuli as a result oftransient and sustainedspaiial attenti-on. MUllrn, H. J., & Rrnnrrr, P. M. A. (1989). Reflexive
and voluntary
Psyc hophysio logy, 33, 13-21.
orienting ofvisual attention: Time course ofactivation and resistance
Etrvrn, M. ( 1999). Facilitatory and inhibitory effects of masked prime
to interruption. Journal of Experimental Psychology: Human perstimuli on motor activation and behavioral performance. Acta psyception & Performance, 15, 3 I 5-330.
chologica, l0l, 293-3 13.
Nerlvllra, K., & MlcxnnrN, M. (1989). Sustainedand transientcomEItr.tsn,M., & ScHr-ncHncxEN,F. (1998). Effects of maskedstimuli on
ponents of focal visual attention. Vision Research. 29. 163l-1647 .
motor activation: Behavioral and electrophysiological evidence.
NErr-1,W. T., Valons, L. A., & Tbmv, K. M. (1995). SelectiveattenJournal of Experimental Psychologt: Human Perception & perfortion and the inhibitory control of cognition. In F. N. Dempster & C. J.
m a n c e . 2 4 .1 7 3 7 - 1 7 4 7 .
Brainerd (Eds.), Interference and inhibition in cognition (pp.207EnIxsEN,B. A., & EnxsrN, C. W. (1974). Effects of noise lettersupon
261). San Diego: Academic Press.
the identification ofa target letter in a visual search. perception &
NsuMrNN, O., & Klorz, W. (1994). Motor responsesto nonreportable,
Psychophys ics, 16, 143 - | 49.
maskedstimuli: Where is the limit of direct parameterspecification? In
EruxsEN,C. W., & HonpunNx, J. E. ( 1973).The extentofprocessins of
C. Umiltd & M. Moskovitch (Eds.), Attention and performance XV:
noise elementsduring selectiveencoding from visual disptays.FerConscious and nonconscious information processing (pp. 123-150).
ception & Psychophysics, 14, 155-160.
Cambridge, MA: MIT Press, Bradford Books.
EruxsrN,C. W., PaN,K., & Borur-n, J. (1993).Attentionaldistribution
PeN, K., & Eruxsnr.r,C. W. (1993). Attentional distribution in the visual
in visual space.Psychological Research,56, 5-13.
field during same-differentjudgments as assessedby responsecomFox, E. ( I 995). Negative priming from ignored distractors in visual sepetition. Perception & Psychophysics, 53, 134-144.
lection: A review. Psychonomic Bulletin & Review, 2, 145-173.
Ptnny, V. H., & Cowrl A. (1984). Retinal ganglion cells that project
I382
SCHLAGHECKEN AND EIMER
to the superior colliculus and pretectum in the macaque monkey.
Neuroscience,12, | 125-l | 37 .
P o s N E RM
, . I . , S N v o t n , C . R . R . , & D a v r o s o N ,B . J . ( 1 9 S 0 ) .A t t e n t i o n
and the detection of signals. Journal of Experimental psychology:
General, 109. | 60-l'74.
R a y n o N o , J . E . , S s n p r n o ,K . L . , & A n N r n , K . M . ( 1 9 9 2 ) .T e m p o r a l
suppressionand visual processingin an RSVP task: An attentional
blink? Journal of Experimental Psychologl,: Human perception &
Performance.18. 849-860.
R n v n o N o ,J . 8 . , S H n p r n oK
, . L . , & A n N E r _ rK
_ ,. M . ( 1 9 9 5 ) .S i m i l a r i t y
determinesthe attentional blink. -/ournal of Experimental psychology:
Human Perception & Performance,2l, 653-662.
ReN4rNcroN,R. W., Jor.rNsloN,J. C., & YaNrrs, S. (1992). Involuntary
attentional capture by abrupt onsets.Perception & Psychophysics,5l,
279-290.
RIcclo, L., & KrnsNEn,K. (1997). The relationship between central
cues and peripheral cues in covert visual orientatlon.perception &
Psychophysics, 59, 885-899.
Rrrsnnx, J. P, KnooN, J. N., & vlN oen WrLor, G. J. (1980). Contrast
sensitivity as a function of position on the retina. Vision Research, 20.
235-241.
ScuLacHecxEN,F., & Eurarn,M. (1997). The influenceof subliminally
presented primes on response preparation. Sprache und Kognition,
16,t66-175.
Snrro,H. G. O. M., MULDER,G., & Mur_oEn,L. J. M. (1990). Selective
responseactivation can begin before stimulus recognition is complete:
A psychophysiologicaland error analysisof continuous flow. Acta
Psychologica,7 4, 169-201.
Ttppsn,S. P.( 1985).The negativepriming effect: Inhibitory priming by
ignored objects. Quarterly Journal o.fExperimental psychology,
37A.571-590.
NOTES
l. The effectivenessofthe masking procedurein preventingthe consciousperceptionofthe primes was testedin numerous forced choice
performance blocks in which participants had to detect the presence of
a prime or identify masked primes presentedeither without (Schlaghecken & Eimer, 1997) or with (Eimer & Schlaghecken,1998) subsequent targets.Detection and identification performances were at chance
level, and participantsconsistentlyreported their inability to respond
discriminatively, suggestingthat the masking procedure was effective in
preventing subjective awarenessof the primes.
2. Experiment I A is a partial replication of Schlaghecken and Eimer
(1997). In this study, positive compatibility effects obtained with peripheral primes were substantially smaller than the corresponding negative compatibility effects obtained with central primes, and 4 out of l0
participants even showed a tendency for negative compatibility effects
with peripheral primes in one or the other SOA condition. Since the experiment was very demanding for the participants (consisting of40
RT blocks and 20 different experimental conditions), we assumethat this
reflected lessthan-perfect eye movement control (note that for primes
presentedat or near fixation, negative compatibility effects are to be expected). Therefore, we felt it necessaryto confirm with a less demanding set-up that the CPA is a reliable phenomenon.
3. Additional forced choice detection blocks for central and peripheral primes were delivered at the end ofboth experiments, including
masked primes l6 msec in duration, as well as primes of longer durations, to prevent participants from switching to random guessing because of their perceived inability to perform as instructed. Detection
performance for central and peripheral l6-msec primes was at chance
level, showing that the masking procedure successfully prevented conscious awarenessofthe primes.
4. The bilateral central targets were arranged vertically, rather than
horizontally (as in Experiment l), in order to counteract an increased
tendencyto move the eyesto the left or the right, revealedin a pilot study,
which is most likely related to the inclusion of unilateral targets in the
left or the right visual field.
5. One ofthe four conditions ofExperiment 2 (central primes, long
mask-targct SOAs) is similar to a previous experiment reported by Eimer
(1999, Experiment l). A negative compatibility effect was obtained for
both target t1pes, providing initial evidence that perceptual repetition
blindness is not responsible for this effect. Since no peripheral primes
were used in this experiment and SOA was not varied, this experiment
did not provide any direct evidence about the relationship between the
physical similarity of the primes and targets and the CPA.
6. Note that this result cannot be explained as an artifact ofeye movement: Ifparticipants had moved their eyes to the cued location, one
should expect a negative compatibility effect for cued prime trials, because masked primes would be at or close to fixation and foveal primes
are known to elicit a negativecompatibility effect at the prime-target interstimulus interval employed in the present experiment (Eimer, 1999;
Eimer & Schlaghecken,1998; Schlaghecken& Eimer, 1991.
(Manuscript received February 9, 1999;
revision accepted for publication December I 5, I 999.)