Functional dissociations in top–down control dependent neuralrepetition priming
Peter Klaver,a,b,d,⁎ Malte Schnaidt,c,d,1 Jürgen Fell,d Jürgen Ruhlmann,e
Christian E. Elger,d and Guillén Fernándezf
aDepartment of Psychology, University of Zurich, SwitzerlandbMR Center, Children's University Hospital Zurich, SwitzerlandcDepartment of Psychiatry, State Hospital Bonn, GermanydDepartment of Epileptology, University Hospital Bonn, GermanyePrivate Institute for Diagnostic and Therapeutic Neuroradiology, Bonn, GermanyfF.C. Donders Center for Cognitive Neuroimaging and Department of Neurology, Radboud University Nijmegen, The Netherlands
Received 17 August 2006; revised 23 October 2006; accepted 2 November 2006Available online 18 December 2006
E-mail address: [email protected] (P. Klaver).1 Malte Schnaidt contributed equally to the study as the first author.Available online on ScienceDirect (www.sciencedirect.com).
A powerful way to investigate verbal representations is tomeasure the facilitation (or “priming”) by the repetition of words orassociated stimuli. Currently, an increasing number of studies haveused decreased neural responses induced by word repetition toinvestigate verbal representations in the brain (Henson, 2003;Schacter et al., 2004). Word repetition priming leads to a reducedblood oxygenation level dependent (BOLD) responses in manybrain areas (Dehaene et al., 1998; Wagner et al., 1998) or smalleramplitudes of event-related potentials (ERP) recorded from scalpelectrodes and depth electrodes within the medial temporal lobe(Halgren et al., 1994; Nobre and McCarthy, 1995; Rugg andDoyle, 1994; Smith et al., 1986). These studies revealed a cascadeof brain areas that show neural priming during the processing ofwords. In the current study we investigate how top–down controlaffects neural correlates of word repetition priming. At the neurallevel, little is known about these mechanisms. We discuss possiblemechanisms after introducing behavioral and neural mechanismsof repetition priming in verbal processing.
Behavioral priming studies essentially contributed to thedevelopment of models on verbal processing. For example,priming studies distinguished between perceptual and conceptualpriming. Perceptual priming is a form of priming that is affected bydifferences in the physical features of the prime and primed stimuli.Conceptual priming is affected by differences in the degree ofsemantic processing of the stimuli and thus is most likely to occurin semantic tasks (Roediger and McDermott, 2003). Such and otherdissociations have led to models of visual word processing. Mostof these models assume three levels of processing that may interactwith each other. A perceptual level extracts visual/perceptualcompounds to letters and pronounceable phonemes. A wordidentification level matches the phonemic or visual word forminformation to lexical information, which determines whether the
1734 P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
word can be interpreted. A conceptual level associates lexicalinformation with semantically associated concepts (Levelt, 1999).
Brain imaging studies identified several brain areas showingneural priming that were associated with this cascade of visualword processing during reading and semantic decision tasks(Henson, 2003; Indefrey and Levelt, 2001; Price, 2000). Forexample, posterior and medial areas in the occipital cortex havebeen associated with early stages of visual perceptual processing(Dehaene et al., 2001). At a later stage of perceptual identification,word compounds may be conjoined to a single word formrepresentation and matched with lexical information. Theseprocesses have been associated with the left occipitotemporal(LOTC) and posterior middle temporal gyrus (pMTG). Reductionsof brain activity in the LOTC, at the border of the posteriorfusiform gyrus and inferior temporal gyrus, may reflect visualword form priming (Cohen et al., 2002) although a strict account ofthis claim has been strongly criticized (Price and Devlin, 2003). Ahigher order function of this area in reading has been suggested inamodal priming studies using auditory and visual stimuli. Primingstudies using a stem-completion task showed within-modality butnot between-modality neural priming in this area (Badgaiyan et al.,2001; Schacter et al., 1999). Similar within-modality neuralpriming effects were found in a study using magnetoencephalo-graphy during a repetition priming task, suggesting amodalprocessing of words in the LOTC (Marinkovic et al., 2003).Several studies showed larger neural activity for concrete than forabstract words in the adjacent left mid-fusiform gyrus (Fiebach andFriederici, 2004; Mellet et al., 1998). This area showed partialoverlap with object processing (Price et al., 2003) and showedincreased activity during mental imagery (D'Esposito et al., 1997).This suggests a common neural representation for pictures andconcrete words although cross-modal repetition priming has notbeen demonstrated in the LOTC. Functional imaging studiesidentified the posterior middle temporal gyrus and angular gyrus(pMTG/AG), mainly in the left hemisphere to be associated withthe recoding of the written word into a lexical representation. Thishas been classically associated with the Wernicke area. Forexample, these areas were active for words, but not for pseudo-words, and showed lexical priming effects (Hagoort et al., 1999;Kotz et al., 2002; Petersen et al., 1990).
Following lexical identification, the lexical information of theword is semantically associated with other concepts. Two brainregions have been associated with these processes, the anteriormiddle temporal gyrus (aMTG) and left inferior prefrontal cortex(LIPFC). The aMTG was found to show a common representationbetween pictures and words (Vandenberghe et al., 1996). TheaMTG was also involved in semantic priming under conditions thatallowed automatic processing, but did not allow for strategicprocessing (Mummery et al., 1999). Mummery and colleaguesargued that semantic priming may occur automatically in theaMTG, whereas non-automatic priming may be related to differentbrain areas. A good candidate for non-automatic semanticprocessing is the LIPFC. This area may be associated withattentional/strategic semantic and mnemonic processing (Fiez,1997; Gabrieli et al., 1998; Paller and Wagner, 2002). This wassupported by lesion studies showing impaired strategic semanticpriming, whereas automatic priming was unimpaired (Hagoort,1997). Imaging studies showed neural priming in the LIPFC afterstudying words in a semantic task (Poldrack et al., 1998), but not ina perceptual task (Demb et al., 1995), suggesting that neuralpriming in the LIPFC is sensitive to the depth of word processing.
This hypothesis particularly accounts for the anterior part of theLIPFC. That area showed neural priming after word repetitionswithin a semantic task (Wagner et al., 2000) for both words andpictures (Wagner et al., 1997), whereas the posterior part of theLIPFC showed priming both when words were previouslypresented in a perceptual and a semantic task. Both anterior andposterior LIPFC have also been associated with attention and top–down processing (Petrides, 2000). For example, recent meta-analyses reported that these areas were associated with attentivebehavior in tasks that require coordinated orientation, reorientingand tagging to relevant information while ignoring potentiallyconflicting information (Derrfuss et al., 2004, 2005). Together,both anterior and posterior LIPFC may play an important role inboth attention and priming, whereas the aMTG may be related toautomatic semantic processing.
The current study particularly focuses on the role of top–downcontrol of word priming. A typical paradigm to investigate top–down control is the Stroop paradigm (Stroop, 1935). In thisparadigm, conflicting information is presented simultaneously, at asimilar location, while subjects direct attention to one aspect ofinformation and ignore the other. Repetition of the attendedinformation has been associated with priming, i.e. the facilitationof a primed stimulus. Repetition of the ignored information as atarget has been associated with reduction of priming (Maxfield,1997; Mulligan and Hornstein, 2000). Positive priming can bereduced if subjects divide attention between words (Kahneman andTreisman, 1984). The general assumption is that dividing attentionresults in less processing and less priming of an item. This effectmay occur at the perceptual level (Maxfield, 1997; Mulligan andHornstein, 2000) and at the semantic or conceptual level ofprocessing (Stone et al., 2000; Stone et al., 1998). Brain imagingstudies suggested that spatial selective attention may be associatedwith the modulation of the neural priming effect (Eger et al., 2004).Under specific task demands, however, no priming occurs afterignoring a word (Glaser and Düngelhoff, 1984; Smith and Magee,1980). Such an effect occurs particularly when pictures aresemantically processed while simultaneously presented words areignored. Several hypotheses have been made to explain this effect.Semantic processing of words may be slower than that of pictures,and consequently the word's semantic information is not yetavailable when a response to pictures is already given (Smith andMagee, 1980). Alternatively, semantic processing of picturesevokes a general inhibition process on words, which preventswords to induce priming (Glaser and Düngelhoff, 1984). Bothhypotheses explain why words are effectively not processed whilepictures are simultaneously processed semantically. It is howevernot clear which neural mechanisms support this behavioral effect.Ignored words may induce no neural priming at all, or words maybe partially processed but suppressed at a certain stage ofprocessing by attention to the picture, so that no behavioral effectis measured.
We used functional MRI to study neural mechanisms ofpriming. Subjects were presented with a series of conjunctions of aword and picture superimposed on each other. In two Stroop tasks,the subjects were instructed to make a natural/man-made decisionfor either the word or the picture. This design addresses the effectof response competition. The words and pictures were eithersemantically compatible (i.e. natural or man-made) or incompatible(one was natural, the other man-made). The level of interference ofthe ignored stimulus has sometimes been measured with the effectof compatibility, i.e. ignored response incompatible pictures may
Fig. 1. An example trial is shown. Pictures and words were differentlycolored in red or blue. In the first task, subjects made a forced choice natural/man-made decision to words that were superimposed onto pictures. In thisexample, the picture is associated with an incompatible response with theword. In the second task, subjects were instructed to make the same decisiontask for the pictures. In this example, the picture is associated with acompatible response with the word. In the third task, new words werepresented as well as words that previously responded to or ignored as wordsor as pictures.
1735P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
delay the processing of the attended word (Glaser and Düngelhoff,1984). The repetition priming task followed the Stroop tasks. Here,brain activity was measured by fMRI while subjects performed thesame task on a series of old words intermixed with new words(NW). Old words were either previously responded to or ignoredand were either semantically compatible or incompatible with thesuperimposed picture. We will name these words “old attendedcompatible words (OACW)”, “old attended incompatible words(OAICW)”, “old ignored compatible words (OICW)” and “oldignored incompatible words (OIICW)”. We expected that pre-viously attended words on a semantic task exhibit reducedresponses in visual word form and semantic processing relatedareas. The question was whether areas showing neural priming forrepeated words reduce priming when attention is directed to thepicture. We also test to whether attention to pictures generally orselectively inhibits processing of words at a specific stage. In otherwords, does specific neural priming occur for previously ignoredwords despite the absence of behavioral priming? Furthermore, wetest whether the semantic compatibility of superimposed words andpictures during study affects the priming of words. If ignoredwords are suppressed while attending pictures, one would expectthat these words show reduced neural priming compared to oldattended words, independently of whether the attended pictureduring study was semantically compatible or incompatible with theword. If, however, ignored words should be semantically orlexically processed, one would expect that processing of theignored word increased by the conflict between semanticallyincongruent superimposed words and pictures and therefore mightinduce increased neural priming in areas related to semantic/lexicalprocessing. In addition, one might hypothesize that the semanticcompatibility of pictures during study affects priming of previouslyattended words. The third question concerned another top–downmechanism in priming, namely cross-modal priming. To investi-gate this issue we show picture names in the repetition task thatwere either initially responded to or ignored as pictures and wereeither semantically compatible or incompatible with superimposedwords. They are called “old attended compatible picture names(OACP)”, “old attended incompatible picture names (OAIP)”, “oldignored compatible picture names (OICP)” and “old ignoredincompatible picture names (OIICP)”. Whereas word priming canbe caused by both perceptual features and semantic properties,picture name priming can only be caused by the commonconceptual properties of the word and the picture. The questionwas whether pictures show cross-modal priming effects in areasthat were involved in the processing of words. We also test whethercross-modal priming is reduced when attention is directed towords. Finally, we test whether priming of picture names isaffected by the semantic compatibility of the superimposed wordspresented during study. If ignored words are not processed duringstudy, priming of previously attended picture names should not beaffected by the semantic compatibility of superimposed wordsduring study, whereas priming of ignored pictures may be affectedby the compatibility of previously attended words.
Methods
Participants
Twenty-four healthy volunteers participated in the experiment(9 female, mean age 27, range 21–37). All participants were righthanded (Edinburgh handedness inventory). All subjects had
normal or corrected-to-normal vision, none had a history ofsignificant neurological disorders, and all gave informed writtenconsent.
Stimuli and procedure
Participants viewed a series of trials in three tasks (see Fig. 1 foran example). In two Stroop tasks, line-drawings of commonobjects (Snodgrass and Vanderwart, 1980) were presented super-imposed on German nouns (250 ms) and were followed by a delay(2000–2800 ms). A central fixation cross was shown during thedelays. The pictures and words were differently colored (red/blue)and were presented on a white background. Subjects responded toeither the words or the pictures by instruction in the twosuccessive Stroop tasks. The repetition priming task immediatelyfollowed the Stroop tasks with a mean delay between first andsecond presentation of about 2 min. Only words were presented(250 ms), which had the same color as in the Stroop tasks andwhich were followed by a delay (2000–2800 ms). The repetitionpriming experiment consisted of new words and words that werepreviously responded to or ignored. Pictures that were previouslyresponded to or ignored were repeated as picture names. In total,200 words were presented during the repetition priming task withequally sized groups of new, old attended words, old ignoredwords, old attended picture names and old ignored picture names.Half of the words were natural, half were man-made. Naturalstimuli were animals (61), fruits or vegetables (23), plants (4) orbody parts (12). Man-made stimuli were clothes (19), tools (38),musical instruments (10) or other non-living objects (33). Tocontrol for the validity of the German picture names, we instructedten different healthy subjects to make a picture naming task toforty living and forty non-living pictures. Picture names wereincluded in the study if they had consistent responses for at least80% of the subjects.
1736 P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
Each task started with an instruction followed by a countdownand two ‘warm-up’ trials that were not included in the behavioraland imaging analysis. In the Stroop tasks, subjects made a forcedchoice natural/man-made decision for either the words or thepictures. Half of the ignored stimuli (words or pictures) wereresponse compatible with the attended stimuli, half wereincompatible. Responses to natural and man-made stimuli wereassigned to the index or little finger, which was counterbalancedacross subjects. The color of the pictures and words was alsocounterbalanced across subjects. The order of the Stroop condi-tions was counterbalanced so that half of the subjects responded towords or pictures that were ignored by the other half of thesubjects. We did not balance the words and pictures due to a lack ofa sufficient number of natural and man-made pictures. To accountfor a possible bias in the priming task for words and picture names,we compared response times on a man-made/natural decision taskfor words (885 ms) and picture names (881 ms) in a pilot study(t<1, ns). There were no significant differences between words andpictures for the word length (range Word: 3–9, Picture name: 3–13), number of syllables (range W: 1–3, P: 1–4) and wordfrequency (range W: 0–1091, P: 0–2830) (all p>0.1).
Critical fMRI data were acquired during the priming task. Theexperiment employed an event-related design and lasted about25 min. The 200 trials in five stimulus categories were pseudo-randomly distributed over the experiment, so that no more thanfour stimuli of the same category were presented in sequence. Theexperiment additionally included 300 null events (1100 ms offixation), which were pseudo-randomly intermixed among thetrials, so that no more than four null-events were presented during asingle inter-stimulus interval. The reason for this design was thatjittering null events increases statistical efficiency when comparingbetween event types (Dale, 1999), whereas increasing delays orrepetitions of the same category may reduce subject's attention.Prior to scanning, participants were informed about the task andpracticed under the supervision of the experimenter. In the scanner,subjects were instructed to respond by pressing a button as quicklyand accurately as possible on a fiber-optic response pad with theright hand. They viewed the stimuli on a backlit projection screenthrough a mirror mounted on the head coil. The task was practicedagain inside the scanner during the anatomical scan. TheExperimental Run-Time System (www.erts.de) was used as astimulus presentation program.
MRI acquisition and analysis
An axial spin-echo planar imaging sequence on a 1.5 T scanner(Siemens Symphony, Erlangen, Germany) was used to measureBOLD contrast. We acquired a series of 381 T2*-weighted scans.The scans were aligned along the AC/PC line. Each whole brainvolume consisted of 25 slices (5 mm with a 0.5 mm gap,3.44×3.44 mm in-plane resolution, field of view=220 mm,repetition time (TR)=2.5 s, echo time (TE)=50 ms). During theStroop tasks, 48 scans were acquired with the same procedure asduring the priming task, so the sound level was comparable duringthe three sessions. However, the design of the Stroop tasks was notoptimized for fMRI data analysis (e.g., null events), and thus thesedata were not analyzed. Anatomical images were acquired using asagittal T1-weighted 3D-FLASH sequence, which was used toidentify the anatomical locations of activations individually (120slices; slice thickness: 1.5 mm without gap; 256×256 matrix;TE=4 ms; TR=11 ms).
Analysis of imaging data was performed using SPM2 (www.fil.ion.ucl.ac.uk/spm). The fMRI data were realigned for move-ment correction and unwarped. To correct for their differentacquisition times, the signal measured in each slice was shiftedrelative to the acquisition time of the middle slice using a sincinterpolation in time. The fMRI data were then normalized to anSPM template with a resampled voxel size of 4×4×4 mm andsmoothed with a Gaussian kernel (full width at half maximum:8 mm). The expected hemodynamic responses at stimulus onsetfor five stimulus categories (new words, old attended and ignoredwords, old attended and old ignored picture names) were modeledby two response functions, which were a canonical hemodynamicresponse function (HRF) and its temporal derivative (Friston et al.,1998). The functions were convolved with the event train ofstimulus onsets to create covariates in a general linear model. Thevector onsets started after two dummy trials in order to let thesubjects accommodate to the task. The first three scans whichcorresponded with the time between the countdown before the taskand the third trial of the experiment were excluded from theanalysis. Only correct responses were modeled. Incorrect responseswere modeled with a regressor of no interest. Parameter estimatesfor each covariate were obtained by maximum-likelihood estima-tion while using a temporal high-pass filter (cut-off 128 s) andmodeling temporal autocorrelation as an AR(1) process. All SPMcomparisons were performed as random effects analyses across 24subjects employing a one-way within-subject ANOVA with fourstimulus categories that were contrasted against new words.Regions of interest (ROIs) with a diameter of 10 mm wereselected on the basis of the coordinates of the visual word formarea (Dehaene et al., 2001), anterior and posterior inferiorprefrontal cortex (Wagner et al., 2000). The ROIs provide asmall number of a priori regional hypotheses. The signal changesof four contrasts (NW-OAW, NW-OIW, NW-OAP and NW-OIP)were extracted from these regions and submitted to a two-wayANOVA with the factors attention (priming attended and ignored)and stimulus type (word repetition priming and picture–picturename priming). In a second analysis we estimated nine contrastswith old stimulus categories that were separated for whether thesewere compatible or incompatible with the superimposed stimulusduring study. Signal changes of eight contrasts entered a three-way ANOVA with the factors stimulus type (word, picture),attention (attended, ignored) and compatibility (compatible,incompatible). We also report clusters of brain activity showingsignificant main effects or interactions after correction for thewhole brain.
Results
Behavioral results
StroopAs can be seen in the left part of Fig. 2, we found an effect of
response compatibility on words. A two-way ANOVA wasperformed which showed a main effect of stimulus type (F(1,23)=35.9, p<0.001) indicating that natural/man-made decisions forpictures were faster than for words. We also found an interactionbetween response compatibility and stimulus type (F(1,23)=6.5,p<0.05). This effect could be explained by a compatibility effect forwords (t(23)=2.3, p<0.05), but not for pictures (t(23)=1.2, ns).These results indicate that word processing was affected by pictureinformation, but not vice versa.
Fig. 2. The bar graphs show mean behavioral performance values with standard errors (n=24). The left column shows response times during the two Strooptasks. The right column shows the priming effect (new–old) for previously ignored or attended words and pictures that were repeated as picture names.
1737P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
PrimingAs seen in the right part of Fig. 2, the priming effect was
substantially influenced by a prior response to the stimulus and bythe type of stimulus. We found only a significant priming effect forOAW (t(23)=3.5, p<0.05). All other categories were not significant(maximum: t(23)=1.3, ns). We found an interaction betweenstimulus type and Stroop condition (F(1,23)=6.0, p<0.05). Thisinteraction could be explained by a larger priming effect for OAWthan for OIW (t(23)=3.4, p<0.01), whereas there was no differencebetween OAP and OIP (t(23)=0.4, ns). To control for the delaybetween the first and second presentation, we tested whethersubjects having the word task first differed in priming from thosethat had the picture task first. There was no significant behavioralpriming effect of order or interaction between task order and primingor attention (all F<1). A three-way ANOVA with the factorsattention, stimulus type and semantic compatibility showed nofurther dissociation. There was no significant effect of semanticcompatibility or interaction with semantic compatibility. Takentogether, attending and responding to pictures seemed to reducerepetition priming for simultaneously presented words. No evidencewas found for a reliable priming effect of picture names, neitherwhen they were previously attended nor ignored.
fMRI effects of attention on repetition priming
Since we were particularly interested in areas related topriming, we defined ROIs on the basis of previous studies thatshow word repetition priming and semantic priming effect forwords. We analyzed three ROIs: the left anterior and posteriorinferior prefrontal gyrus (aLIPFC (x=−43, y=32, z=15), pLIPFC(−43, 8, 34)), left fusiform gyrus (left occipital temporal cortex,LOTC (−46, −57, −20)). In the LOTC we found a significanteffect of attention (F(1,23)=4.5, p=0.045), but not of stimulustype or an interaction between stimulus type and attention. As canbe seen in Fig. 3, this indicated that attended stimuli showedincreased priming as compared to ignored stimuli. In the aLIPFCwe found trends to significance for attention (F(1,23)=3.1,p=0.09), stimulus type (F(1,23)=3.4, p=0.08) and the interactionbetween stimulus type and attention (F(1,23)=3.5, p=0.07).Comparing priming for old attended with old ignored stimuli inthe aLIPFC resulted in a significant difference between OAW andOIW (t(23)=2.6, p=0.017) but not between OAP with OIP (t(23)=0.3, ns). There was also a significant difference between OAW andOAP (t(23)=2.7, p=0.013). There were no main effects orinteractions for the pLIPFC (all F<3, ns) (Fig. 3). Thus, attention
during study affected priming of both words and picture names inthe LOTC, whereas the aLIPFC showed only an effect of attentionon word repetition priming. We also performed the same factorialanalysis on the whole brain and found no significant clustersshowing interactions between attention and stimulus type. Wordsinduced increased activity compared to pictures in the parahippo-campal gyrus (BA 30, local max Z=3.7, p=0.002, x/y/z=−4/−40/−4) and extrastriate cortex (BA 19, Z=4.4, p=0.001, x/y/z=12/−88/24). A trend to significance was found in the anterior middleoccipital gyrus for the comparison between priming of previouslyattended words and ignored words (OAW-OIW: BA 19 Z=4.0,p=0.065, x/y/z=−52/−60/−8). This area largely overlapped withthe ROI of the LOTC.
fMRI effects of attention and semantic compatibility on neuralpriming
Next, we tested whether semantic compatibility of super-imposed words and pictures during study affected subsequentpriming. An ANOVA with the factors stimulus type, attention andsemantic compatibility was submitted to the three ROIs. In theLOTC we found an interaction between attention and semanticcompatibility (F(1,23)=5.9, p=0.024) as well as trends tosignificance for the effect of attention (F(1,23)=4.2, p=0.052)and compatibility (F(1,23)=4.0, p=0.056). As can be seen in thelower part of Fig. 3, priming seemed to be reduced whensemantically incompatible stimuli were superimposed during studyor when stimuli were previously ignored (attended compatible–incompatible: t(23)=3.2, p=0.004; ignored comp–incomp: ns). Asignificantthree-wayinteractionindicatedthatthiseffectwasdifferentfor words and pictures (F(1,23)=5.3, p=0.03). There was a trend tosignificance for the effect of attention on words (attention words: F(1,23)=3.7, p=0.065). In contrast, picture to picture name primingwas affected by both the semantic compatibility of the wordand attention (interaction attention by compatibility: F(1,23)=12.1, p=0.002). Comparing compatible and incompatible stimuliwe found only a significant effect between compatible andincompatible attended pictures (t(23)=4.9, p<0.001), but not forignored pictures or attended or ignored words (all t<2, ns).Hence, the neural priming effect of previously attended pictures onpicture names was reduced in the LOTC if the superimposed wordduring study was semantically incompatible with the picture.
The two frontal regions aLIPFC and pLIPFC showed a similarpattern of results, but different from the LOTC. We foundinteractions between attention and semantic compatibility (aLIPFC:
Fig. 3. Event-related fMRI activation rendered upon a standard brain shows significant top–down modulations of neural repetition priming for words in tworegions of interest (left column) that show main effects of attention (“red”), and between words that were previously ignored or attended (“yellow”). Right upperpart shows top–down effects on neural priming in a whole-head analysis for significant clusters showing main effects of compatibility (“purple”), interactionsbetween compatibility and attention (“green”) and interactions between stimulus type, attention and compatibility (“blue”). For visualization, effects were color-coded at p<0.001. The lower part shows bar-plots of mean signal changes due to neural priming (%) with standard error bars at three ROIs (left occipito-temporal cortex, LOTC; anterior and posterior left inferior prefrontal cortex, aLIPFC/pLIPFC) for four contrasts (OAW, OIW, OAP and OIP). Below showssignal changes for eight contrasts including stimulus compatibility in the LOTC and aLIPFC. On the right side are shown signal changes in local maxima in thewhole-head analysis for areas showing significant results in the main effect and interactions listed above.
1738 P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
F(1,23)=7.0, p=0.014; pLIPFC: F(1,23)=6.4, p=0.019). Theseinteractions can be explained by the stronger effect of compatibilityon previously ignored stimuli as compared to previously attended
stimuli. The priming effect was larger for previously ignored stimulithat were incompatible with the superimposed stimulus during studythan for those that were compatible with superimposed stimuli
1739P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
(ignored comp vs. incomp: aLIPFC t(23)=−2.3, p=0.032, pLIPFC t(23)=−2.4, p=0.027). No such effect occurred when stimuli werepreviously attended (attended comp vs. incomp, ns). Together,priming in the LOTC reduced when the stimuli were previouslyignored and incompatible with the superimposed stimulus duringstudy. This was particularly the case for previously attendedpictures. Both anterior and posterior LIPFC showed no effect ofcompatibility on previously attended stimuli but showed an increasein priming if stimuli were previously ignored and incompatible.
A whole brain analysis showed both main effects of semanticcompatibility, interactions between attention and compatibility onpriming and three-way interactions between attention, compat-ibility and stimulus type. Detailed results are listed in Table 1 andFig. 3. Semantic compatibility affected neural priming in theposterior cortex, particularly in extrastriate regions of the middleoccipital gyrus and lingual gyrus (BA 18/19). Here, we foundlarger priming effects for compatible stimuli as compared toincompatible stimuli during study. We found significant interactionsbetween attention and semantic compatibility in the left posteriorinferior temporal cortex (including LOTC), left anterior andposterior inferior/middle prefrontal cortex (including aLIPFC andpLIPFC) as well as the right anterior middle frontal gyrus (BA 9/46),medial prefrontal gyrus and anterior cingulate gyrus (BA 9/32),right cerebellum, and posterior cingulate gyrus and precuneus (BA30/31). All these areas, except for the left inferior/middleprefrontal gyrus, showed decreased neural priming for previouslyattended stimuli with incompatible as compared with compatiblesuperimposed stimuli during study. Three-way interactions werefound in the occipital cortex including the middle occipital gyrus,lingual gyrus (BA 18 and 19) and primary visual cortex (BA 17).These areas showed no neural priming for words, or effects ofattention or compatibility on word repetition priming. However,these areas showed an interaction between attention and compa-tibility on priming of picture names (Fig. 3). This interaction
Table 1
Main effect:compatibility
Frontal lobe Z pR medial FG (−4 32 36) BA 8/9Ant cingulate G (−4 24 44) BA 32R inf/middle FG (48 32 20) BA 9/46L middle FG/precentral G (−36 4 56) BA 6/9
Occipital/Temporal lobeL fusiform G (−32 −36 −24) BA 20L middle occipital G (−52 −64 −16) BA 37L post cingulate G (−4 −52 12) BA 30Precuneus (4 −68 16) BA 31L middle occipital G (−32 −96 0) BA 18/19R middle occipital G (24 −100 4) BA 18/19L lingual G (−12 −72 0) BA 18/19Primary visual cortex (4 −88 4) BA 17R lingual G (0 −88 −12) BA 18 4.6 <0.001L middle occipital G (−48 −80 −8) BA 19 4.1 0.011
CerebellumR cerebellum (40 −72 −32)
Functional MRI results show attention and semantic compatibility related effects osignificance (*) are shown (p<0.05). The local maxima of each cluster are listedcluster. Brodmann areas (BA) are depicted for each cluster. Abbreviations: L/R is
indicated that previously attended pictures induced reduced primingon picture names if they were presented with semanticallyincompatible words during study.
Discussion
The present study investigated the neural mechanisms under-lying top–down control of word priming. Behavioral datareplicated studies showing that words shown at fixation during asemantic decision task on superimposed pictures significantlyreduced the behavioral priming effect. We found no evidence forcross-modal priming and no effect of attentional modulation oncross-modal priming. Furthermore, semantic compatibility did notsignificantly affect priming of words or picture names. Thesebehavioral priming data support the hypothesis that ignored wordsare not processed while making a semantic response to super-imposed pictures (Glaser, 1992; Glaser and Düngelhoff, 1984).Brain activity showed more complex effects of top–downmodulation of priming. First, neural priming for old ignoredwords was reduced compared to old attended words in the LOTCand aLIPFC. In addition, we found effects of cross-modal primingand interactions between cross-modal priming and attention. Wealso found evidence that semantic compatibility of superimposedwords and pictures affected both priming of words and picturenames. We will discuss these findings in the context of currentliterature.
The first question was whether neural repetition priming wasaffected by attention. Imaging data support the hypothesis thatattention increased neural priming. We found priming effects in theLOTC that increased for previously attended as compared toignored stimuli. These effects were independent of whether a wordor a picture name was primed. These data extend previous findingson attention dependent neural priming (Eger et al., 2004). Thatstudy reported reduced neural priming effects in object processing
Interaction:attention bycompatibility
Interaction:stimulus by attentionby compatibility
Z p k Z4.2 0.0013.83.9 0.0463.4 0.024
4.3 <0.0013.383.83 <0.0013.57
4.5 0.0014.1 0.0084.2 0.075*
3.75 0.023
n neural repetition priming. Significant clusters or areas showing a trend toas x/y/z coordinates in MNI space, as well as distant areas within the sameleft/right, prefix ant/post=anterior/posterior, FG=frontal gyrus, G=gyrus.
1740 P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
sensitive fusiform gyrus if spatial attention was directed away fromobjects. There were two main differences between both studies. Inour study we used words and found effects of attention in a morelateral and word form processing sensitive area (LOTC or visualword form area). Secondly, in contrast to Eger and colleagues, weused superimposed words and pictures so that attention affectedstimulus processing even though both words and pictures were inthe center of fixation. Thus, our findings support and extend thatneural priming is modulated by selective attention.
The second question was if priming of ignored words wasgenerally inhibited by a semantic decision to pictures, as proposedby Glaser (1992), and whether this would result in a general absenceof neural priming effect or whether ignored words were partiallyprocessed and resulted in selective neural priming. We found small,but not significant priming for ignored words, so that statisticalpower may be too small to detect priming. Imaging data, however,suggest that ignored words partially induce neural priming. First, wefound significant differences in neural priming of the anterior LIPFCfor ignored words depending on whether these were semanticallycompatible or incompatible with superimposed pictures duringstudy. This finding suggests that semantically incompatible wordsinduce more processing during study than compatible words,possibly because of a semantic interference with superimposedpictures. Behaviorally we found no evidence for such a semanticinterference during the study task since responses to previouslyignored words did not depend on the semantic compatibility ofpictures during study. These data cannot be explained by responseswitches between study and test. Dobbins and colleagues showedthat neural priming of objects reduced in the fusiform gyrus andLIPFC when they were repeated after a response switch (Dobbins etal., 2004). They suggested that priming was strongly affected by therepetition of response to the stimulus. In the current study, ignoredwords that were semantically incompatible with superimposedpictures also required a different response on the first and secondword presentation. One might hypothesize that such responseswitches reduce priming. However, we found an increased neuralpriming effect for response incompatible stimuli. Thus, the currentfinding may be explained by a response or semantic competitioneffect during study but not by response switching between study andtest for semantically incompatible ignored stimuli.
Secondly, we found differences in neural priming by previouslyattended pictures depending on the semantic compatibility with thewords. Picture names showed reduced priming when they werepresented during study with semantically incompatible word ascompared to semantically compatible words. Thus, the words musthave been processed to a semantic level. One might alsohypothesize that these effects relate to response switchingdependent priming. As noted above, Dobbins and colleagues(2004) reported that response switches reduce priming effects. Inthe current study, one might hypothesize that response informationof incompatible ignored words competes with responses topictures. If the response of a word would be preferred, this wouldthen result in a response switch situation for the picture. This is,however, an unlikely hypothesis since behavioral data indicate thatpictorial information affects responses to words during the Strooptask, but not vice versa. Thus, word information of the ignoredword reduces picture to picture name priming on the basis ofsemantic interference. Together, there is ample evidence thatignored words induce neural priming at a semantic level despite thestrong top–down modulation when subjects make a semanticdecision task to pictures.
The current findings seem to contradict findings showing thatwords are not processed during processing of pictures (Rees et al.,1999). An explanation for differences in these results may berelated to the working load conditions during encoding. Rees andcolleagues suggested that words are not processed at all whilepicture processing was difficult in a visual matching task. In ourcase the perceptual or processing load in the picture task waslower, so that ignored words might have been processed up to acertain level.
The third issue was whether pictures may prime picture namessemantically and whether these priming effects depend onattention. We found no significant behavioral priming effect ofpictures to picture names, which is in line with studies showingthat picture to picture name priming is smaller than word repetitionpriming also if words and pictures are presented separately (Dursoand Johnson, 1979). However, we found a main effect of attentionon neural priming in the LOTC, which may be equivalent to thevisual word form area and an interaction between stimulus typeand attention in the aLIPFC. This suggests that picture names wereprimed similarly as words, at least in the LOTC. As far as we knowcross-modal picture to word effects have not been reported in theLOTC, but these results are in line with the top–down modulationhypothesis for the LOTC (Buckner et al., 2000). They are also inline with findings showing semantic priming effects for objects inthe fusiform gyrus (Simons et al., 2003) and support the criticismthat LOTC processes only visual word forms (Price and Devlin,2003). In summary, these data support the hypothesis that theLOTC plays an important role in attention dependent priming forconceptual information.
Finally, we were interested if cross-modal priming effects wereaffected by top–down modulations of semantic compatibilityduring study. Several brain areas showed similar top–downmodulations on neural priming of picture names as of words.Interactions between attention and compatibility in several areasincluding the left fusiform gyrus and frontal cortex showedreduced neural priming responses for previously attended picturenames and words when superimposed stimuli during study weresemantically incompatible. Only the aLIPFC showed an increasedresponse for ignored words and pictures when superimposedstimuli during study were semantically incompatible. These resultsstrongly supported that cross-modal priming effects were modu-lated by attention and compatibility in a similar way. As far as weknow this is the first study to show such effects. Several studieshave reported common representations of pictures and words (Priceet al., 2003; Vandenberghe et al., 1996) or effects of concretenessin the fusiform gyrus (Fiebach and Friederici, 2004; Mellet et al.,1998). The finding that priming of both normal and mirroredobjects are similarly affected by spatial attention in this area alsosuggests top–down modulations of a higher-order representation(Eger et al., 2004). Thus, the current data suggest that cross-modalpriming, or common representations between words and pictures,at least partially underlie the same top–down control mechanisms.
The current study furthermore shows strong effects of semanticinterference on neural priming. The effects of interference onpriming occurred particularly in the posterior cortex. We foundmain effects of compatibility, as well as interactions betweenattention and compatibility and three-way interactions betweenstimulus type, attention and compatibility. These results stronglysupport the role of semantic interference in the top–down control ofpriming. Interestingly, semantic compatibility affected differencesbetween old and new stimuli even in areas that are not particularly
1741P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
sensitive to word priming, including several striate and extrastriatecortex. The data particularly suggest that top–down mechanismsaffect neural priming related to pictures. A possible explanation forthese findings may be related to the fact that picture names aredirectly associated with previously presented pictures. This associa-tion may have activated a top–down process in a similar way asduring a mental imagery task and by means of an increase in neuralactivity in perceptual processing areas during repetition (Kosslyn etal., 1995). We also found that neural priming in these areas wasreduced for picture names when previously attended pictures werepresented with semantically incompatible words, whereas neuralpriming of words was unaffected by the semantic compatibility ofsuperimposed pictures. An explanation for this effect may be thatsemantic interference of words intervenes with semantic processingof pictures. Repetition of pictures may then have induced lesspriming in perceptual processing areas.
Several other reasons have been discussed that modulate neuralpriming. For example, a prominent top–down effect on priming isnegative neural priming. Negative priming represents the increaseof response latencies after repetition of an ignored stimulus orconceptually related stimulus (Damian, 2000; Tipper and Driver,1988). It is usually observed when ignored stimuli are repeated inthe presence of a distractor and reduces when no distractinginformation is present (Allport et al., 1985; Lowe, 1979), but mayalso occur when no distracting information is present (Fox, 1995;Moore, 1994; Neill et al., 1994; Yee, 1991). Negative priming mayreflect the cost of retrieving previously ignored information afterthe active inhibition of stimulus information (Neill et al., 1992).This may be associated with increased activity in the rightdorsolateral prefrontal cortex (Egner and Hirsch, 2005). Increasedactivity after repetition of ignored stimuli has also been reported inmedial parts of the inferior temporal lobe (Gazzaley et al., 2005;Vuilleumier et al., 2005). In the current study, no evidence wasfound for negative neural priming of ignored words, neither infrontal areas, nor in perceptual processing areas. There was also noevidence that negative neural priming coexisted with (orneutralized) positive neural priming when words showed nobehavioral priming effect. There was, however, a difference withother brain imaging studies showing negative priming (Egner andHirsch, 2005; Steel et al., 2001; Vuilleumier et al., 2005). In thepriming task, we presented only the ignored item withoutdistracting picture. Several behavioral studies showed thatnegative priming depends on the presence of a distractor duringprobe stimulus presentation (Allport et al., 1985; Lowe, 1979;Tipper and Cranston, 1985). For example, if only color patches(without the word) were presented following a conventionalcolor Stroop task, positive priming instead of negative primingwas found (Lowe, 1979). This suggests that priming can bereversed depending on whether a distractor is present or not. Atleast, studies showing negative priming without a distractorcreated an expectation that distractors might occur (May et al.,1995; Neill et al., 1994). In the current study we showed nodistractors at the probe stimulus and created no expectation thata distractor might occur. Thus, in the context of those studies,the current imaging data are in line with the hypothesis that nonegative neural priming for ignored stimuli occurs when nodistractor is presented or expected, although caution must betaken since increases and decreases in neural activity may occurin the same brain areas (Gazzaley et al., 2005). Thesemodulations in neural activity may be related to positive andnegative priming. In such a case, the simultaneous occurrence of
positive and negative priming may be mistaken for a modulationof positive priming.
Another aspect may modulate priming, namely stimulusfamiliarity. Previous studies reported that unfamiliar stimuli showincreased neural responses upon repetition, whereas familiarstimuli show decreased responses (Grill-Spector et al., 2006;Henson et al., 2000). In the current study, we use familiar stimuli ofwords and pictures. These stimuli did not differ between categoriessince these were counterbalanced across subjects. Thus, this factormay not have affected the modulation of priming in the currentstudy.
Taken together, the present study reveals multiple mechanismsof top–down control on repetition priming. We report that neuralpriming is affected by attention, semantic compatibility and cross-modal priming. We also show interactions between these top–down mechanisms on neural priming. In particular, neuralpriming of pictures to picture name was similarly affected byattention and semantic compatibility as word repetition priming,and ignored words induced neural priming effects at a semanticlevel, despite the absence of significant behavioral priming. Thesedata thus extend previous studies showing modulations of neuralpriming.
References
Allport, D.A., Tipper, S.P., Chmiel, N.R.J., 1985. Perceptual integration andpostcategorical filtering. In: Posner, M.I., Marin, O.S.M. (Eds.),Attention and Performance XI. Erlbam, Hillsdale, NJ, pp. 107–132.
Badgaiyan, R.D., Schacter, D.L., Alpert, N.M., 2001. Priming within andacross modalities: exploring the nature of rCBF increases and decreases.NeuroImage 13 (2), 272–282.
Buckner, R.L., Koutstaal, W., Schacter, D.L., Rosen, B.R., 2000. FunctionalMRI evidence for a role of frontal and inferior temporal cortex in amodalcomponents of priming. Brain 123, 620–640.
Cohen, L., Lehericy, S., Chochon, F., Lemer, C., Rivaud, S., Dehaene, S.,2002. Language-specific tuning of visual cortex functional properties ofthe visual word form area. Brain 125, 1054–1069.
Dehaene, S., Naccache, L., Le Clec'H, G., Koechlin, E., Mueller, M.,Dehaene-Lambertz, G., van de Moortele, P.F., Le Bihan, D., 1998.Imaging unconscious semantic priming. Nature 395 (6702), 597–600.
Dehaene, S., Naccache, L., Cohen, L., Le Bihan, D., Mangin, J.F., Poline, J.B., Riviere, D., 2001. Cerebral mechanisms of word masking andunconscious repetition priming. Nat. Neurosci. 4 (7), 752–758.
Demb, J.B., Desmond, J.E., Wagner, A.D., Vaidya, C.J., Glover, G.H.,Gabrieli, J.D.E., 1995. Semantic encoding and retrieval in the leftinferior prefrontal cortex—a functional MRI study of task difficulty andprocess specificity. J. Neurosci. 15 (9), 5870–5878.
Derrfuss, J., Brass, M., von Cramon, D.Y., 2004. Cognitive control in theposterior frontolateral cortex: evidence from common activations in taskcoordination, interference control, and working memory. NeuroImage23 (2), 604–612.
Derrfuss, J., Brass, M., Neumann, J., von Cramon, D.Y., 2005. Involvementof the inferior frontal junction in cognitive control: meta-analyses ofswitching and Stroop studies. Hum. Brain Mapp. 25 (1), 22–34.
D'Esposito, M., Detre, J.A., Aguirre, G.K., Stallcup, M., Alsop, D.C.,Tippet, L.J., Farah, M.J., 1997. A functional MRI study of mental imagegeneration. Neuropsychologia 35 (5), 725–730.
Dobbins, I.G., Schnyer, D.M., Verfaellie, M., Schacter, D.L., 2004. Corticalactivity reductions during repetition priming can result from rapidresponse learning. Nature 428 (6980), 316–319.
1742 P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
Durso, F.T., Johnson, M.K., 1979. Facilitation in naming and categorizingrepeated pictures and words. J. Exp. Psychol. Hum. Learn. Mem. vol. 5,449–459.
Eger, E., Henson, R.N.A., Driver, J., Dolan, R.J., 2004. BOLD repetitiondecreases in object-responsive ventral visual areas depend on spatialattention. J. Neurophysiol. 92 (2), 1241–1247.
Egner, T., Hirsch, J., 2005. Where memory meets attention: neural substratesof negative priming. J. Cogn. Neurosci. 17 (11), 1774–1784.
Fiebach, C.J., Friederici, A.D., 2004. Processing concrete words: fMRIevidence against a specific right-hemisphere involvement. Neuropsy-chologia 42 (1), 62–70.
Fiez, J.A., 1997. Phonology, semantics, and the role of the left inferiorprefrontal cortex. Hum. Brain Mapp. 5, 79–83.
Fox, E., 1995. Negative priming from ignored distractors in visualselection—a review. Psychon. Bull. Rev. 2 (2), 145–173.
Gabrieli, J.D., Poldrack, R.A., Desmond, J.E., 1998. The role of left pre-frontal cortex in language and memory. Proc. Natl. Acad. Sci. U. S. A.95 (3), 906–913.
Gazzaley, A., Cooney, J.W., McEvoy, K., Knight, R.T., D'Esposito, M.,2005. Top–down enhancement and suppression of the magnitude andspeed of neural activity. J. Cogn. Neurosci. 17 (3), 507–517.
Glaser, W.R., 1992. Picture naming. Cognition 42, 61–105.Glaser, W.R., Düngelhoff, F.J., 1984. The time course of picture–word
interference. J. Exp. Psychol. Hum. Percept. Perform. 10 (5),640–654.
Grill-Spector, K., Henson, R., Martin, A., 2006. Repetition and the brain:neural models of stimulus-specific effects. Trends Cogn. Sci. 10,14–23.
Hagoort, P., 1997. Semantic priming in Broca's aphasics at a short SOA: nosupport for an automatic access deficit. Brain Lang. 56, 287–300.
Hagoort, P., Indefrey, P., Brown, C., Herzog, H., Steinmetz, H., Seitz, R.J.,1999. The neural circuitry involved in the reading of German words andpseudowords: a PET study. J. Cogn. Neurosci. 11, 383–398.
Halgren, E., Baudena, P., Heit, G., Clarke, M., Marinkovic, K., 1994.Spatiotemporal stages in face and word processing 1: depth recordedpotentials in the human occipital, temporal and parietal lobes. J. Physiol.(Paris) 88 (1), 1–50.
Henson, R., Shallice, T., Dolan, R., 2000. Neuroimaging evidence fordissociable forms of repetition priming. Science 287 (5456), 1269–1272.
Indefrey, P., Levelt, W., 2001. The neural correlates of language production.The Cognitive Neurosciences. MIT Press.
Kahneman, D., Treisman, A.M., 1984. In: Parasuraman, R., Davies, D.R.(Eds.), Varieties of Attention. Academic Press, London, pp. 29–61.
Kosslyn, S.M., Thompson, W.L., Kim, I.J., Alpert, N.M., 1995. Topo-graphical representations of mental images in primary visual cortex.Nature 378, 496–498.
Kotz, S.A., Cappa, S.F., von Cramon, D.Y., Friederici, A.D., 2002.Modulation of the lexical–semantic network by auditory semanticpriming: an event-related functional MRI study. NeuroImage 17 (4),1761–1772.
Levelt, W.J.M., 1999. Models of word production. Trends Cogn. Sci. 3,223–232.
Lowe, D.G., 1979. Strategies, context, and the mechanism of response-inhibition. Mem. Cogn. 7 (5), 382–389.
Marinkovic, K., Dhond, R.P., Dale, A.M., Glessner, M., Carr, V., Halgren,E., 2003. Spatiotemporal dynamics of modality-specific and supramodalword processing. Neuron 38 (3), 487–497.
Maxfield, L., 1997. Attention and semantic priming: a review of prime taskeffects. Conscious. Cogn. 6 (2–3), 204–218.
May, C.P., Kane, M.J., Hasher, L., 1995. Determinants of negative priming.Psychol. Bull. 118 (1), 35–54.
Mellet, E., Tzourio, N., Denis, M., Mazoyer, B., 1998. Cortical anatomy of
mental imagery of concrete nouns based on their dictionary definition.NeuroReport 9, 803–808.
Moore, C.M., 1994. Negative priming depends on probe-trial conflict—Where has all the inhibition gone. Percept. Psychophys. 56 (2), 133–147.
Nobre, A.C., McCarthy, G., 1995. Language-related field potentials in theanterior–medial temporal lobe: II. Effects of word type and semanticpriming. J. Neurosci. 15 (2), 1090–1098.
Paller, K.A., Wagner, A.D., 2002. Observing the transformation ofexperience into memory. Trends Cogn. Sci. 6 (2), 93–102.
Petersen, S.E., Fox, P.T., Snyder, A.Z., Raichle, M.E., 1990. Activation ofextrastriate and frontal cortical areas by visual words and word-likestimuli. Science 1041–1044.
Petrides, M., 2000. Mapping prefrontal cortex systems for the control ofcognition. In: Toga, A.W., Mazziotta, J.C. (Eds.), Brain Mapping: TheSystems. Academic Press, pp. 159–176.
Poldrack, R., Wagner, A.D., Prull, M.W., Desmond, J.E., Glover, G.H.,Gabrieli, J.D.E., 1998. Functional specialization for semantic andphonological processing in left inferior prefrontal cortex. NeuroImage10, 15–35.
Price, C.J., 2000. The anatomy of language: contributions from functionalneuroimaging. J. Anat. 197, 335–359.
Price, C.J., Devlin, J.T., 2003. The myth of the visual word form area.NeuroImage 19 (3), 473–481.
Price, C.J., Noppeney, U., Phillips, J., Devlin, J.T., 2003. How is thefusiform gyrus related to category-specificity? Cogn. Neuropsychol. 20(3–6), 561–574.
Roediger, H.L., McDermott, K.B., 2003. Implicit memory in normalsubjects. In: Boller, F., Grafman, J. (Eds.), Handbook of Neuropsychol-ogy. Elsevier, Amsterdam, pp. 63–161.
Rugg, M.D., Doyle, M.C., 1994. Event-related potentials and stimulusrepetition in direct and indirect tests of memory. In: Heinze, H.J., Münte,T.F., Mangun, G.R. (Eds.), Cognitive Electrophysiology. Birkhauser,Boston, pp. 124–148.
Schacter, D.L., Badgaiyan, R.D., Alpert, N.M., 1999. Visual word stemcompletion priming within and across modalities: a PET study.NeuroReport 10 (10), 2061–2065.
Simons, J.S., Koutstaal, W., Prince, S., Wagner, A.D., Schacter, D.L., 2003.Neural mechanisms of visual object priming: evidence for perceptual andsemantic distinctions in fusiform cortex. NeuroImage 19 (3), 613–626.
Smith, M.C., Magee, L.E., 1980. Tracing the time course of picture-wordprocessing. J. Exp. Psychol. Gen. 109 (4), 373–392.
Smith, M.E., Stapleton, J.M., Halgren, E., 1986. Human medial temporallobe potentials evoked in memory and language tasks. Electroencepha-logr. Clin. Neurophysiol. 63, 145–159.
Snodgrass, J.G., Vanderwart, M., 1980. A standardized set of 260 pictures:norms for name agreement, image agreement, familiarity, and visualcomplexity. J. Exp. Psychol. Hum. Learn. Mem. 6 (2), 174–215.
Steel, C., Haworth, E.J., Peters, E., Hemsley, D.R., Sharma, T., Gray, J.A.,Pickering, A., Gregory, L., Simmons, A., Bullmore, E.T., Williams, S.C.,2001. Neuroimaging correlates of negative priming. NeuroReport 12(16), 3619–3624.
1743P. Klaver et al. / NeuroImage 34 (2007) 1733–1743
Stone, M., Ladd, S.L., Vaidya, C.J., Gabrieli, J.D.E., 1998. Word-identification priming for ignored and attended words. Conscious.Cogn. 7 (2), 238–258.
Stone, M., Ladd, S.L., Gabrieli, J.D.E., 2000. The role of selective attentionin perceptual and affective priming. Am. J. Psychol. 113 (3), 341–358.
Stroop, J.R., 1935. Studies of interference in serial verbal reactions. J. Exp.Psychol. 18, 643–662.
Tipper, S.P., Cranston, M., 1985. Selective attention and priming: inhibitoryand facilitatory effects of ignored primes. Q. J. Exp. Psychol. 37A,591–611.
Tipper, S.P., Driver, J., 1988. Negative priming between pictures and wordsin a selective attention task: evidence for semantic processing of ignoredstimuli. Mem. Cogn. 16 (1), 64–70.
Vandenberghe, R., Price, C., Wise, R., Josephs, O., Frackowiak, R.S., 1996.Functional anatomy of a common semantic system for words andpictures. Nature 383 (6597), 254–256.
Selective attention modulates neural substrates of repetition priming and“implicit” visual memory: suppressions and enhancements revealed byfMRI. J. Cogn. Neurosci. 17 (8), 1245–1260.
Wagner, A.D., Desmond, J.E., Demb, J.B., Glover, G.H., Gabrieli, J.D.E.,1997. Semantic repetition priming for verbal and pictorial knowledge: afunctional MRI study of left inferior prefrontal cortex. J. Cogn.Neurosci. 9 (6), 714–726.
Wagner, A.D., Schacter, D.L., Rotte, M., Koutstaal, W., Maril, A., Dale,A.M., Rosen, B.R., Buckner, R.L., 1998. Building memories: remem-bering and forgetting of verbal experiences as predicted by brain activity.Science 281 (5380), 1188–1191.
Wagner, A.D., Koutstaal, W., Maril, A., Schacter, D.L., Buckner, R.L., 2000.Task-specific repetition priming in left inferior prefrontal cortex. Cereb.Cortex 10 (12), 1176–1184.
Yee, P.L., 1991. Semantic inhibition of ignored words during a figureclassification task. Q. J. Exp. Psychol., Sect. A Hum. Exp. Psychol. 43 (1),127–153.
