This article was originally published in a journal published byElsevier, and the attached copy is provided by Elsevier for the

author’s benefit and for the benefit of the author’s institution, fornon-commercial research and educational use including without

limitation use in instruction at your institution, sending it to specificcolleagues that you know, and providing a copy to your institution’s

administrator.

All other uses, reproduction and distribution, including withoutlimitation commercial reprints, selling or licensing copies or access,

or posting on open internet sites, your personal or institution’swebsite or repository, are prohibited. For exceptions, permission

may be sought for such use through Elsevier’s permissions site at:

http://www.elsevier.com/locate/permissionusematerial

Autho

r's

pers

onal

co

py

Research Report

Electrophysiological correlates of familiarity in recognitionmemory and exclusion tasks

N.C. Bridson, C.S. Fraser, J.E. Herron, E.L. Wilding⁎

School of Psychology, Cardiff University, Cardiff, CF10 3AT, Wales, UK

A R T I C L E I N F O A B S T R A C T

Article history:Accepted 13 July 2006Available online 28 August 2006

ERPs were acquired in the test phases of three memory experiments, where three classes ofword were presented. These were: (i) words encountered in a prior study phase (studiedwords), (ii) words presented at test for the first time (new words), and (iii) newwords repeatedafter a lag of 7–9 intervening words (repeated test words). In experiments 1 and 2, participantswere asked to respond on one key to studied words and on another to new as well as torepeated test words. In experiment 3, a binary response was again required, but in this caserepeated test and studied words were assigned to the same key. In each experiment, theprincipal focus for analysis was on the differences between the ERPs atmid-frontal electrodelocations from 300 to 500ms post-stimulus that were associated with incorrect responses tostudied words (misses) and correct responses to new words. It has been proposed thatrelatively greater positivity for studied than for new words at this locus reflects the greaterfamiliarity of studied than of unstudied words. ERPs elicited by misses were reliably morepositive-going than those elicited by correct rejections in experiments 1 and 2 only. Thesefindings support the link between this modulation of the electrical record and familiarity inso far as the designs of the experiments lead to the prediction that the average level offamiliarity associated with misses should be higher in the first two experiments than in thethird. In combination with findings in other studies, these data support dual-processaccounts of recognition memory.

© 2006 Elsevier B.V. All rights reserved.

Keywords:FamiliarityRecollectionEvent-related potentialsExclusion taskEpisodic memory

1. Introduction

ERP old/new effects are differences between the scalp-recorded neural activity associated with correct memoryjudgments to old (previously studied) and new test stimuli(for reviews, see Curran et al., in press; Friedman and Johnson,2000; Mecklinger, 2000; Rugg et al., 2002). In recognitionmemory tasks where only old/new recognition judgmentsare required, two old/new effects are often observed. Theeffect with the longer history is the left-parietal ERP old/neweffect, which, as the name suggests, is largest over left-parietal

scalp—at least for verbal stimuli. The effect comprises arelatively greater positivity for old than for new items. Thispositivity onsets around 500 ms post-stimulus and lastsbetween 200 and 800 ms, the duration depending upon taskdemands and task performance (Donaldson and Rugg, 1999;Wilding and Sharpe, 2003).

This ERP old/new effect has been linked with the process ofrecollection–recovery of qualitative information about a prioroccurrence (Yonelinas, 2002). Briefly, the evidence that sup-ports this view includes the following: (1) the amplitude of theeffect is correlated positively with the amount of contextual

B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

⁎ Corresponding author. Fax: +44 29 20874858.E-mail address: wildinge@cardiff.ac.uk (E.L. Wilding).

0006-8993/$ – see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.brainres.2006.07.095

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ l oca te /b ra in res

Autho

r's

pers

onal

co

py

information that is recovered (Wilding, 2000; Wilding andRugg, 1996), (2) the effect is larger for correct old judgmentsassociated with Remember (R) rather than Know (K) judg-ments (Duzel et al., 1997; Smith, 1993), (3) the effect isattenuated or absent in patients where selective brain damageor disease has led to an impairment restricted primarily torecollection (Duzel et al., 2001; Rugg et al., 1991; Smith andHalgren, 1989; Tendolkar et al., 1999).

The second old/new effect observed in recognitionmemorystudies is evident between 300 and 500 ms post-stimulus, alsocomprises a relatively greater positivity for old than for newtest items, but is most prominent at mid-frontal electrodelocations (e.g. F3, Fz, F4)1. The effect is on occasions referred toas the FN400 (for review, see Curran et al., in press), but isreferred to here as the mid-frontal ERP old/new effect. Theeffect has been linked with familiarity (e.g. Mecklinger, 2000),but the data germane to this issue is not as clear cut as thatsupporting the link between the left-parietal ERP old/neweffect and recollection.

Across a number of experiments, a link between the mid-frontal ERP old/new effect and familiarity has been proposedon the basis of the pattern of effects associated with certaintypes of incorrect responses, namely false alarms to luresresembling closely studied items. For example, in the studydue to Curran (2000), participants studied words in singular orplural forms (e.g., frogs, lake) and were tested with copy cues(frogs, lake), new words (capitol) and lures where plurality wasthe opposite of that at study (lakes, frog). On the basis ofprevious behavioural findings (Hintzman and Curran, 1994),Curran reasoned that, in the absence of recollection of studymaterial, lures should seem as familiar as studied words andshould attract more old responses than should new words.The behavioral data was in line with previous findings, andthe putative indices of familiarity were of equivalent magni-tude for old responses to old words and to lures, while theindex of recollection was larger for copy cues. These data areconsistent with the link between the mid-frontal ERP old/neweffect and familiarity described above, and qualitativelysimilar findings have been reported in other studies inwhich different stimulus materials and different similarlure/old item relationships have been employed (Curran etal., 2002; Nessler et al., 2001; Penney et al., 2001).

For other findings, however, the behaviour of the mid-frontal old/new effect cannot be accommodated straightfor-wardly by a familiarity interpretation. These include the factthat the mid-frontal ERP old/new effect is insensitive to somebehavioural manipulations that are assumed to influence theavailability of familiarity, in particular depth of processing(Rugg et al., 1998), full versus divided attention, and the levelsof confidence associated with recognition memory judgments(Curran, 2004). In addition, Yovel and Paller (2004) failed toobtain reliable mid-frontal ERP old/new effects for pictures offaces, despite behavioural indications that the critical stimuli

were familiar to participants in the context of the experiment.In other studies where non-verbal stimuli have been used,however, reliable mid-frontal ERP old/new effects have beenobserved (Nessler et al., 2005; Penney et al., 2001).

The failure to observe changes in the mid-frontal ERP old/new effect in a manner that supports a familiarity accountmay, at least in some of these studies, be a consequence of alack of statistical power (Azimian-Faridani and Wilding, 2006;Curran, 2004). The results, none the less, remain at oddswith afamiliarity interpretation, and from the perspective of modelsof recognition memory this issue is an important one becauseof the continuing debate about the extent to which data inrecognition memory tasks and close variants can be accom-modated within single or dual-process frameworks (e.g. Dunn,2004; Rotello et al., 2004; Wixted and Stretch, 2004; Yonelinas,2001b; Yonelinas et al., 1996). Given the relatively strongevidence linking the left-parietal ERP old/new effect torecollection (see above), the identification of a separablemodulation that behaves as an index of familiarity wouldprovide support for dual-process accounts of recognitionmemory.

Motivated by these considerations, the three studiesdescribed here were designed in order to assess the familiarityaccount of the mid-frontal ERP old/new effect in a somewhatdifferent way to those described above, using variants of theexclusion task procedure (Jacoby, 1991; Jacoby and Kelly, 1992).In the typical incarnation of this task, participants study itemsinitially in one of two contexts. In a subsequent test phase,items from both contexts are presented alongside new items.Participants are informed that they must make one key pressto items from one of the two study contexts, and another keypress to genuinely new test items as well as to items from theother study context.

In a variant of this task, test items from two separatecontexts comprise items encountered in a prior study phase,and new test items that are repeated after a designated lag.This design was introduced by Jennings and Jacoby (1997), andhas been employed by Dywan and colleagues in a series ofstudies that were motivated by different questions than thoseat issue here (Dywan et al., 1998, 2001, 2002). We return to thecorrespondence between their findings and those describedhere in the Discussion.

The reason for using this task in order to assess thecorrespondence between the mid-frontal ERP old/new effectand familiarity is because consideration of the likely basis formemory judgments permits predictions concerning how themid-frontal ERP old/new effect should behave if indeed itindexes familiarity. The critical category here is incorrectresponses to studied items, and the principal insight is that inthis task high levels of familiarity in the absence of recollec-tion are not necessarily a reliable basis for task judgments.

Consider the task as described above, the instructions toparticipants being to respond on one key to studiedwords, andon the other to words presented at test only (hereafter newwords and repeated testwords). Given a short repetition lag, therepeated test items are likely to be highly familiar, so highlevels of familiarity are a good cue for identifying a repeateditem. Consider, however, studied items. If recollection andfamiliarity are independent bases for task judgments (Jacoby,1998; Jacoby et al., 1997), then some studied items will be

1 When an average reference is employed, positive-going ERPold/new effects at mid-frontal electrodes are accompanied bypolarity reversed effects distributed over posterior scalp sites.These old/new effects are commonly restricted to mid-frontalscalp locations when a linked-mastoid reference is used, as in thecurrent experiment (for review, see Curran et al., in press).

150 B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

py

familiar but not recollected. For these items, moderate to highlevels of familiarity for studied words that are not recollectedwould not be a good basis for task judgments, assuming thatmany repeated test words are highly familiar because of theshort lag between first and second presentations. As a result,familiar but not recollected studiedwords can attract incorrectresponses despite relatively high levels of familiarity.

The prediction that follows from this argument is that, ifthe mid-frontal ERP old/new effect indexes familiarity, then areliable mid-frontal ERP old/new effect should be obtained forincorrect judge to studied items. This prediction was tested intwo of the three experiments described below. The first twoexperiments were designed as described above, the principaldifference between them being the use of different encodingtasks. In Experiment 1, participants were asked to read eachstudy word aloud. In Experiment 2, participants were asked togenerate and say aloud a word that rhymed with each studyword.

In the third experiment, the same encoding task as inExperiment 2 was employed. The critical manipulation was attest. Participantswere again asked tomake a binary judgment,but in this experiment were asked to respond on the same keyto all items presented twice (studied items plus repeated testitems) and on another key to items presented for the first timeat test. In this case, familiarity is a good basis for taskjudgments, because both repeated and studied items aremapped to the same response. Under these task conditions,therefore, incorrect responses to both classes of old (studiedand repeated) test items should be made only on the basis ofrelatively low levels of familiarity. It follows from this thatlittle evidence of a mid-frontal ERP old/new effect should beobtained in this experiment if indeed the mid-frontal ERP old/new effect indexes familiarity.

2. Results

2.1. Behavioural data

Table 1 shows the behavioural data for all three experiments.In each case, the table displays the likelihood of a correctresponse to studied, new and repeated test words. Thereaction times for each experiment for correct responses arealso shown in Table 1. For the first two experiments, two

measures of discrimination (Pr: Snodgrass and Corwin, 1988)were calculated, the first to determine discrimination betweenstudied and repeated test words. Themean Pr valueswere 0.53and 0.56 in experiments 1 and 2, respectively. These Pr valueswere reliably greater than 0, indicating that participants wereable to discriminate between these two classes of test item(Exp. 1: t(15)=10.44, p<0.001; Exp. 2: t(15)=11.61, p<0.001).Values of Pr were also computed in order to assess discrimina-tion between studied and new words. The estimates wereagain reliably greater than 0 in both cases (Exp. 1: Pr=0.62,t(15)=20.71, Exp. 2: Pr=0.60, t(15)=14.63, p<0.001). These fourmeasures of discrimination were submitted to ANOVA,incorporating the factor of Experiment, and revealed nosignificant differences. For Experiment 3, the values of Pr forstudied and repeated words were both reliably greater than 0(studied: Pr=0.62, t(15)=21.37, p<0.001, repeated: Pr=0.81,t(15)=22.38, p<0.001), and reliably different from each other(t(15)=12.88, p<0.001). There were no reliable differencesbetween the Pr measures involving studied words for Experi-ment 3 and for experiments 1 and 2, as assessed by t-tests.

The RTs for correct responses to each class of test wordwere subjected to ANOVA, with Experiment included as afactor. The analysis revealed a significant interaction betweenexperiment and category (F(2.9, 66.2)=4.83, p<0.01). Follow-upanalyses were completed separately for each experiment, ineach case comprising all possible paired contrasts (t-tests)between the three response categories. For Experiment 1, theanalysis revealed no reliable differences. For experiments 2and 3, the analysis revealed significant differences betweenthe RTs associated with correct responses to studied andrepeated test words (Exp. 2: t(15)=5.36, p<0.001, Exp. 3 t(15)=6.47, p<0.001) as well as between RTs for new words andrepeated test words (t(15)=3.43, p<0.005 and t(15)=3.14,p<.01). In all cases, the faster responses were to the repeatedtest words.

In a second set of analyses that mirror the ERP analysesdescribed below, the RTs associated with incorrect responsesto studied words (misses2) were contrasted with thoseassociated with correct rejections. The mean RTs for misseswere 1089, 1286 and 1366 ms in experiments 1, 2 and 3,respectively. In each experiment, paired t-tests revealed thatthe RTs for correct rejections were faster than those formisses (Exp. 1: t(15)=2.65, p<0.05; Exp. 2: t(15)=3.15, p<0.01;Exp. 3: t(15)=4.27, p<0.001).

2.2. ERP data

ERPs could in principle be formed for six response categoriesin each experiment: correct and incorrect responses tostudied, repeated test and new words. In each experiment,the low number of incorrect responses to repeated test andnew words precluded formation of reliable averaged ERPs for

Table 1 – Probabilities of correct responses and reactiontimes (RT) to studied, new and repeated test words inexperiments 1, 2 and 3 (SD in brackets)

Studied New Repeated

Exp. 1p(correct) 0.69 (0.13) 0.93 (0.06) 0.83 (0.13)RT 908 (395) 876 (353) 875 (348)

Exp. 2p(correct) 0.67 (0.18) 0.94 (0.03) 0.90 (0.08)RT 1137 (691) 1038 (626) 903 (448)

Exp. 3p(correct) 0.73 (0.08) 0.89 (0.10) 0.92 (0.05)RT 1033 (302) 1096 (592) 821 (208)

2 Incorrect responses to studied words in Experiments 1 and 2might equally well be referred to as false alarms, given that anincorrect response to a studied word means a response on the keydesignated for correct responses to new and repeated words. Thisis not the case in Experiment 3, however, and for consistencyacross experiments the term miss is employed throughout thismanuscript when referring to incorrect responses to studiedwords.

151B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

py

the majority of participants. ERPs were therefore analysed forfour categories: correct responses to all classes of test words,and incorrect responses to studied words. The mean trialnumbers and ranges per participant contributing to the ERPsfor each of these four response categories in each experimentare shown in the Appendix.

The ERPs were subjected to a series of directed analysesfocused on the sites at whichmid-frontal and left-parietal ERPold/new effects are typically observed (Mecklinger, 2000;Nessler et al., 2005; Rugg et al., 1998; Wilding and Rugg, 1997;Wilding and Sharpe, 2003). In the following sections, theanalyses for all three experiments for the mid-frontal effectsare described first, followed by the left-parietal ERP old/neweffects. In each case, the analyses were first completed for theERP old/new effects for correct responses, followed by theanalyses for incorrect responses. For experiments 2 and 3, theanalyses of the mid-frontal old/new effects incorporated datafrom five electrode locations: F3, F1, Fz, F2, F4. For Experiment1, the sites were F3, Fz and F4 (see Experimental procedures).These analyses were completed over the 300–500 ms timewindow and included the factors of response category andsite. For all three experiments, the analyses of the left-parietalERP old/new effects incorporated the data from P5, P3, P4 andP6 and were completed over the 500–700 ms time window,including the factors of category, hemisphere and site.

2.3. Mid-frontal ERP old/new effects

Fig. 1 shows the ERPs elicited by correct responses to studiedwords, new and repeated test words at electrode sites F3, Fzand F4. The figure shows that, from approximately 300 ms

onwards, the ERPs elicited by studied and repeated test wordsare more positive-going than those elicited by new words. Fig.2 shows data from the same electrode locations for incorrectresponses to studied words (misses) and for correct responsesto newwords. The figure shows that, in the critical 300–500mstime window, the ERPs elicited by misses are more positive-going than those elicited by correct rejections, and that thisrelative positivity is moremarked in experiments 1 and 2 thanin Experiment 3.

The initial analyses of the mid-frontal ERP old/new effectsincluded mean amplitudes for correct responses to studiedwords, new words and repeated test words. In each experi-ment, this analysis revealed a main effect of category only(Exp. 1: F(1.9, 29.2)=18.56, p<0.001; Exp. 2: F(1.9, 28.7)=22.82,p<0.001; Exp. 3: F(1.7, 26.1)=5.62, p<0.01). These analyseswere followed up by all possible paired contrasts within eachexperiment. The outcomes of these analyses are shown in theupper half of Table 2, and demonstrate that, while notdiffering in magnitude, there are reliable old/new effects forstudied words as well as for repeated test words. The onlyinteraction with site in these analyses was obtained inExperiment 1, reflecting the fact that the mid-frontal ERPold/new effect for studied words is largest at Fz.

The outcomes of the critical contrast between misses andcorrect rejections are shown in Table 3, which shows that themean amplitudes associated with misses were reliably morepositive-going than those associated with correct rejections inexperiments 1 and 2 only. Fig. 2 illustrates this critical findingacross the three experiments. In a further analysis, the meanamplitudes associatedwith correct rejections andwithmisseswere contrasted across experiments. The amplitudes entering

Fig. 1 – Grand average ERPs elicited by correct responses to studiedwords aswell as new and repeated testwords at sites F3, Fzand F4 in Experiments 1, 2 and 3.

152 B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

pythe analysis were those from the 300–500 ms time window forsites F3, Fz and F4. This analysis revealed no reliable effects.The scalp distributions of the ERP old/new effects for studiedand repeated test words attracting correct judgments areshown in Fig. 3. The figure also shows the scalp distribution ofthe differences betweenmisses and correct rejections over thesame epoch for each experiment. The distribution of the ERPold/new effect for misses is somewhat left-lateralised inExperiment 2 relative to that in Experiment 1. A direct contrastbetween these effects involving all scalp locations common tothese experiments and computed over difference scoresobtained by subtracting mean amplitudes for correct rejec-

tions from those for misses did not, however, reveal aninteraction between experiment and scalp site.

2.4. Left-parietal ERP old/new effects

Fig. 4 shows the data from five parietal electrode locations forcorrect responses to studied and repeated words (hits andrepeated hits, respectively) and correct rejections. The figureshows that while the two classes of hits differ only minimallyfrom each other, both are more positive-going than the ERPselicited by correct rejections. This relative positivity peaksaround 600 ms post-stimulus and only in Experiment 3 is

Table 2 – Outcomes of the paired contrasts between the ERPmean amplitudes associatedwith correct responses to studiedwords (Hits), repeated test words (RHits) and correct rejections (CRs) for the 300–500 and 500–700 ms time windows inexperiments 1, 2 and 3

Exp. 1 Exp. 2 Exp. 3

Hit/CR RHit/CR Hit/RHit Hit/CR RHit/CR Hit/RHit Hit/CR RHit/CR Hit/RHit

300–500 msRC (1,15) 28.24*** 23.07*** ns 35.51*** 38.21*** ns 8.16* 5.37* nsRC×ST (2,30) 5.95** ns ns ns ns ns ns ns ns

0.98

500–700 msRC (1,15) 31.07*** 86.10*** ns 66.04*** 47.48*** ns 3.71• 26.42*** 7.12*RC×ST (1,15) 20.20*** ns ns 11.21** 7.98* ns ns ns 6.21*RC×HM (1,15) ns ns ns ns ns ns ns ns nsRC×HM×ST (1,15) ns ns ns 3.43• 5.41* ns ns ns ns

RC=response category, HM=hemisphere, ST=site. *p<0.05, **p<0.01, ***p<0.001, •p<0.1, epsilon values shown below respective F-values.

Fig. 2 – Grand average ERPs elicited by correct responses to new words and incorrect responses to studied words inExperiments 1, 2 and 3. Electrode locations as for Fig. 1.

153B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

pythere a strong suggestion that these positive-going differencesare larger at left than at right hemisphere posterior scalplocations.

In keeping with the strategy adopted for the analyses ofthe mid-frontal ERP old/new effects, the initial analysisincluded mean amplitudes for correct responses to studiedwords, new words and repeated test words. In each experi-ment, this analysis revealed a main effect of category (Exp. 1:F(1.4, 21.2)=25.40, p<0.001; Exp. 2: F(1.8, 27.0)=38.81, p<0.001;Exp. 3: F(1.9, 28.7)=11.65, p<0.001). For experiments 1 and 2there were also interactions between this factor and site (Exp.1: F(1.7, 25.3)=6.04, p<0.01; Exp. 2: F(1.7, 25.2)=7.49, p<0.01).These global analyses were followed up by all possible pairedcontrasts, and the outcomes of these, displayed in the lowerportion of Table 2, showed no reliable differences between theERP old/new effects for studied and repeated test words inexperiments 1 and 2, with reliably more positive-going effectsfor repeated test words in Experiment 3. ERPs evoked by bothclasses of old words were reliably more positive-going thanthose evoked by new words in each experiment, with theexception of the studied versus correct rejection contrast inExperiment 3, where the effect approached significance(p=0.07). The interactions between category and site in allcases reflect the fact that the relatively greater positivity forold words is largest at sites closest to the midline (P3/P4). Thethree-way interaction between category, hemisphere and sitethat was obtained in Experiment 2 for the paired contrastbetween repeated test words and new words arose becausethe greater relative positivity for repeated test words is largestat P4. The contrasts between the ERPs evoked by misses andby correct rejections over this time window revealed noreliable effects in any of the three experiments. The ERPwaveforms elicited by misses and correct rejections atparietal electrode locations are shown in Fig. 5, while Fig. 6shows the scalp distributions of the ERP old/new effects overthe 500–700 ms time window for studied and repeated testwords attracting correct judgments in each experiment.

3. Discussion

In all three experiments, participants were able to discrimi-nate between the three critical classes of test word. The timingparameters and stimulus characteristics on test trials wereidentical for all three experiments. Experiments 1 and 2differed only in the encoding task. In experiments 1 and 2,the response required to studied words was on a different keyto that shared by repeated test and new words. Experiment 3

included the same encoding task as Experiment 2, butparticipants were asked to respond on the same key to allwords presented for the second time in the experiment.

The motivation for these studies was to assess the extentto which the mid-frontal ERP old/new effect indexes famil-iarity. The rationale for experiments 1 and 2 was that, in bothexperiments, moderate to high levels of familiarity for studiedwords that were not recollected would not be a good basis fortask judgments. The reason for this claim was the assump-tion that repeated test words would be highly familiarbecause of the short lag between first and second presenta-tion. As a result, familiar but not recollected studied wordsmight attract incorrect responses despite relatively highlevels of familiarity.

It follows from this argument that if the mid-frontal ERPold/new effect indexes familiarity then the effect should beevident in both of these experiments. For Experiment 3,however, the response requirements – respond ‘old’ to allitems presented for the second time in the task – meant thatitem familiarity was a good diagnostic for binary testjudgments, hence incorrect responses to studied items shouldbe associated on average with relatively lower levels offamiliarity in this experiment.

The data are consistent with this account, as reliable mid-frontal ERP old/new effects were obtained for misses only inexperiments 1 and 2, as Fig. 2 shows. The contrasts at parietalsites between misses and correct rejections also revealed no

Fig. 3 – Scalp distributions of the ERP old/new effects forcorrect responses to studied words (Hits) and repeated testwords (rHits) over the 300–500 ms epoch in Experiments 1,2 and 3. Also shown are the scalp distributions of thedifferences between the ERPs elicited by misses and correctrejections. In each case, the spline maps were computed onthe bases of mean amplitudes obtained by subtracting thoseassociated with correct rejections from those associated withhits, repeated hits and new test words, respectively. Theupper and lower limit amplitude values under each mapcorrespond to the progression from red through to blue oneach map.

Table 3 – Outcomes of the paired contrasts betweenmean amplitudes associated with misses and withcorrect rejections over the 300–500 ms window inexperiments 1, 2 and 3

Exp. 1 Exp. 2 Exp. 3

RC (1,15) 10.50** 5.22* nsRC×ST (2,30) ns 3.19• ns

0.47

All nomenclature as for Table 2.

154 B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

pyFig. 4 – Grand average ERPs elicited by correct responses to studiedwords aswell as new and repeated testwords at sites P6, P5,Pz, P3 and P4 in Experiments 1, 2 and 3.

Fig. 5 – Grand average ERPs elicited by correct responses to new words and incorrect responses to studied words inExperiments 1, 2 and 3. Electrode locations as for Fig. 4.

155B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

py

reliable differences in any of the three experiments. This isconsistent with prior findings (Neville et al., 1986; Wilding andRugg, 1996), and also with the view that incorrect responses tostudied words were not associated with substantive levels ofrecollection, as long as it is assumed that activity over thisscalp region is an index of the extent to which recollection isengaged. In all experiments, moreover, there were reliablemid-frontal ERP old/new effects for all classes of old wordattracting correct judgments. These findings converge withthose reported elsewhere in establishing a link between themid-frontal ERP old/new effect and familiarity, and they do soin tasks where for the first time to our knowledge there is noresponse confound: correct responses to new words andincorrect responses to studied words were made on thesame key in experiments 1 and 2.

Whatmay be a complementaryway inwhich to conceive ofthese data is that the familiarity criterion that a studied itemfell below in order to be rejected was lower in Experiment 3than in the other two experiments. The marginally higherfalse alarm and hit rates in Experiment 3 than in the other twoexperiments is consistent with this view, although theambiguity regarding how participants will treat familiar old(studied as well as repeated) test items that are not recollectedmeans that it is not straightforward to estimate accuratelycriterion in the exclusion tasks, nor indeed to assume that asingle criterion model is an appropriate model for this task.Regardless, these observations relate to an alternative meansof framing an account of the mid-frontal old/new effect andfamiliarity, rather than a challenge to the central premise.

Stronger evidence in support of this claim, however, wouldhave stemmed from a reliable effect in the direct contrast ofthe magnitudes of the mid-frontal ERP old/new effects acrossexperiments. Despite the fact that this contrast did notachieve significance, the results are none the less consistentwith the pre-experimental hypotheses. It is also notable that,while not statistically significant, there was a small mid-frontal ERP old/new effect for misses in Experiment 3. In so faras familiarity is a graded process (Yonelinas, 2002; Yonelinaset al., 1996), this small effect is in line with a link between thisERP effect and familiarity since, according to a graded accountof the familiarity signal, on average old items judged to be new

should be more familiar than new items attracting a newresponse. To our knowledge, there are only two studies to datein which it has been shown convincingly that the mid-frontalERP old/new effect indexes familiarity in a graded fashion(Azimian-Faridani and Wilding, 2006; Woodruff et al., 2006).This is important as, in the absence of these findings, theliterature to date is consistent with the view that the mid-frontal ERP old/new effect is an index of familiarity thatbehaves in an all or none fashion. If this account were correctthen it would limit considerably the extent to which it may bepossible to employ changes in the mid-frontal ERP old/neweffect as a functional tool in studies of human long-termmemory (although see Azimian-Faridani and Wilding, 2004).

The pattern of findings reported here is also relevant to thesuggestion that, rather than indexing familiarity, the mid-frontal ERP old/new effect indexes conceptual priming.According to this alternative account, the greater relativepositivity observed for old than for new test items, and forclasses of items such as repeated lures (e.g. Curran, 1999), is aconsequence of the fact that these items are associated with agreater degree of conceptual priming than are new test items,by virtue of their correspondence with items presented atstudy.

It is not clear how this account explains the markedlysmaller amplitude differences between misses and correctrejections at anterior electrodes in Experiment 3 than inexperiments 1 and 2, nor why themid-frontal old/new effect issensitive to the confidence with which recognition memoryjudgments are associated (Azimian-Faridani and Wilding,2006; Woodruff et al., 2006). In so far as conceptual primingand familiarity are independent, these data points challenge apriming account, according to which there is no reason whythemagnitude of the greater relative positivity formisses thanfor correct rejections should vary across experiments (forfurther related comments, see Azimian-Faridani and Wilding,2006; Woodruff et al., 2006). Whether the mid-frontal effectcan act as an index of familiarity for some kinds of non-verbalstimuli, however, remains to be established (Yovel and Paller,2004; Nessler et al., 2005; Curran et al., in press).

In combination with findings in other ERP studies ofrecognitionmemory, therefore, the data reported here supportstrongly dual-process accounts of recognition memory. Thisclaim is based upon the fact that in a number of previousstudies, the left-parietal ERP old/new effect has been linkedwith recollection (see Introduction), so the identification of adistinct scalp modulation with different antecedents arguesstrongly that two functionally distinct processes can con-tribute to recognition memory judgments. In the majority ofERP data sets that provide support for dual-process accounts,moreover, a qualitatively similar pattern has been obtained ineach case: changes in the amplitude of the parietal effect inthe absence of changes across conditions in the mid-frontalold/new effect. The somewhat different pattern here thuspermits an important degree of generalisation. The data donot speak, however, to the question of the relative merits ofdifferent dual-process accounts (Rotello et al., 2004; Yonelinas,2001a, 2002).

Another aspect of the data that warrants comment is theabsence of statistical evidence for left-lateralised parietallydistributed ERP old/new effects in all three studies. Inspection

Fig. 6 – Scalp distributions of the ERP old/new effects forcorrect responses to studied words (Hits) and repeated testwords (rHits) over the 500–700 ms epoch in Experiments 1,2 and 3. All other nomenclature as for Fig. 3.

156 B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

py

of Fig. 3 suggests that there is a degree of lateralisation of theERP old/new effects in Experiment 3, but the relevantstatistical analyses did not reveal interactions involvingcondition and hemisphere (see Table 2). The absence of anasymmetry in the differences between the ERPs evoked byold and new words is arguably more surprising for studiedwords than for repeated test words. This is because incontinuous recognition memory tasks, where participantsencounter a single list of items and some repeat after avariable number of intervening items, parietally distributedERP old/new effects tend to show little if any hemisphereasymmetry (e.g. Rugg and Nagy, 1989; Rugg et al., 1997; Swickand Knight, 1997). To the extent that the repeated test wordsin this task are comparable to repeated words in a standardcontinuous recognition memory task, the absence of asym-metric effects in these experiments is consistent with theprior literature, even if it does not help with an explanationas to why the degree of lateralisation of parietal ERP old/neweffects varies across continuous and study-test recognitionmemory tasks.

From this perspective, the surprising aspects of the data inthese experiments are the parietal ERP old/new effects forstudiedwords. There is no immediate explanation for this nullfinding in the current data, although the absence of stronglylateralised parietally distributed old/new effects has beenreported in other studies (see, for example Van Petten et al.,2000; Wilding et al., 2005; Yovel and Paller, 2004).

The pattern of parietal old/new effects across the threestudies also differs somewhat, with the effects for studied andrepeated test words diverging in Experiment 3 only, where theeffect is larger for repeated test words. One possible reason forthis divergence is the fact that, in experiments 1 and 2 only,correct judgments to new and repeated test words share thesame response option. This means that an unknown propor-tion of “correct” judgments to repeated test words were itemsthatwere forgotten. In so far as forgotten itemselicit little if anyparietal positivity relative to correct rejections in the 500–700 ms time post-stimulus time period, as Fig. 5 shows, theimpact of this on the ERPs is some degree of attenuation of themagnitudeof theparietalold/neweffect for repeatedtestwordscompared to the effect for studied words. Another considera-tion germane to this aspect of the data is the extent to whichparticipants prioritised recollection as a basis for respondingacross experiments, and, within experiments 1 and 2, theextent to which recollection of information associated withstudied and repeated test words was prioritised.

This final observation is apposite because of the disparitybetween the findings reported here and those in relatedstudies. The designs of the experiments described here arevery similar to those employed by Dywan and colleagues in aseries of studies where one focus was on the relationshipsbetween the parietal ERP old/new effects elicited by correctjudgments to studied and repeated test words (Dywan et al.,1998, 2001, 2002).

In each of these three publications3, at least one group ofparticipants was given the same task instructions as those

provided to participants in experiments 1 and 2 describedabove. ERPs at anterior sites in the 300–500 ms time windowwere more positive-going for studied and repeated test words,with a tendency to be more positive-going in the former case.These mid-frontal old/new effects were not, however, ana-lysed directly. The most marked disparities between thefindings of Dywan et al. and those reported here are atposterior-parietal scalp locations. In contrast to the findingsin experiments 1 and 2, parietally distributed ERP old/neweffects in their studies were markedly and reliably larger forcorrect responses to studied words than for correct responsesto repeated test words. In fact, in all three publications, there islittle evidence for a relatively greater positivity for correctresponses to repeated test words in comparison to correctrejections at parietal locations in the 500–700ms timewindow.

The absence or attenuation of parietal ERP old/new effectsfor some classes of old item relative to others has beeninterpreted as evidence for selective attention to task-relevantmaterial only (Dywan et al., 1998, 2001, 2002), or as evidencethat participants can control, at least under some circum-stances, what information will be recollected during a task inorder to complete the task efficiently (Dzulkifli and Wilding,2005; Dzulkifli et al., 2006; Herron and Rugg, 2003a; Hornbergeret al., 2004). These two interpretations are not necessarilymutually exclusive (Herron and Rugg, 2003b), but for presentpurposes the important point is the disparity between thefindings reported here and those due to Dywan and colleagues(Dywan et al., 1998, 2001, 2002).

The differences across studies are unlikely to be due toencoding task, as Dywan et al. required participants simply toread words aloud, the same encoding task that was employedin Experiment 1. In the studies due to Dywan et al., the lagbetween first and second presentations of repeated test wordswas fixed at 6 items, whereas in experiments 1, 2 and 3 the lagvaried between 7 and 9 items. One possibility, therefore, is thatthe greater predictability of the occurrence of repeated testwords contributed to the disparate findings. While thisaccount cannot be ruled out entirely, it seems somewhatunlikely when considering the relatively complex demands ofthe retrieval task, and the fact that in none of the studies wereparticipants informed of the repetition lag.

A second account of the disparities across studies stemsfrom the fact that, in experiments 1 and 2, the ratio of studiedwords to repeated test words was 2:1, contrasting with the 1:1ratio in the studies of Dywan et al. The P300 (P3b) potential issensitive to the relative probabilities of classes of stimuli, aswell as their task-relevance (Donchin and Coles, 1988). TheP300 is typically larger for task-relevant as well as for lowerprobability classesof stimuli, and this component is commonlylargest at centro-parietal scalp sites, with little if any hemi-sphere lateralisation (Horst et al., 1980; Squires et al., 1975).

In light of these observations, this probability disparityacross studies offers to explain the differences between theERP findings. By this account, the greater relative positivity forrepeated test words relative to new words in experiments 1and 2 but not in the work of Dywan et al. is simply aconsequence of a larger P300 elicited by this class of stimuli inexperiments 1 and 2 (see data for Pz shown in Fig. 3). The lowerprobability of occurrence in these experiments, moreover,might have encouraged participants to adopt a different

3 Dywan and colleagues have reported findings on this task foryoung as well as older adults. The description in the text here isfor the data from the young participants only.

157B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

py

strategy for completing the task, focusing to a greater extenton the repeated test words in experiments 1 and 2 than in thestudies of Dywan et al. To the extent that this strategy resultedin the repeated test words being task-relevant, again theprediction would be that this class of stimuli should elicit asizable P300.

It is not possible to distinguish between these accounts onthe basis of the current data, and while they are not theprincipal focus in this paper, these considerations areimportant, because of the growing literature in which it hasbeen claimed that ERPs can be employed as a means ofindexing control over episodic retrieval processing (Dzulkifliand Wilding, 2005; Dzulkifli et al., 2006). Understanding theboundary conditions under which this can be achieved is afundamental part of that research (Rugg et al., 2002).

In summary, reliable mid-frontal ERP old/new effects forincorrect responses were obtained only when the taskdemands rendered familiarity an unreliable basis for criticaltask judgments. These findings support the claim that themid-frontal ERP old/new effect indexes familiarity, at leastwhenwords are employed as test stimuli. In combinationwithfindings from other studies, moreover, the results providestrong support for dual-process accounts of recognitionmemory.

4. Experimental procedures

All three experiments are described jointly due to thesimilarities between their designs.

4.1. Participants

4.1.1. Experiment 119 participants (average age=20, 15 female) completed thetask. The data from three participants (two females) wasdiscarded prior to analysis. For two of these, the reason forrejection was excessive EOG artefact (for criteria see below).The remaining participant did not make sufficient incorrectresponses to studied items. The average age of the remainingparticipants was 20 years (range 18 to 24).

4.1.2. Experiment 224 right-handed people (average age=21, 20 female) completedthe task. The data from eight participants (seven female) wasdiscarded prior to analysis. For two of these the reason forrejection was that they did not make sufficient incorrectresponses to studied items. The remaining participants wereexcluded due to excessive EOG artefact. The average age of theremaining participants was 21 years (range 18 to 30).

4.1.3. Experiment 322 participants (average age=23, 13 female) completed thetask. Thedata fromsix participants (five female)wasdiscardedprior to analysis. For one of these the reason for rejectionwas afailure to demonstrate an ability to distinguishold (studied andrepeated test items) from new items. One participant did notmake sufficient incorrect responses to studied items. For theremainder, there was excessive EOG artefact. The average ageof the remaining participants was 22 years (range 19 to 30).

All participants were right-handed, and no participantswere taking neuroleptic medication at the time of testing orhad a reported history of mental illness. All were paid at therate of £7.50/h and gave informed consent prior to commen-cing the experiments.

4.1.4. DesignFor all three experiments, the stimuli comprised 280 criticalwords from the MRC psycholinguistic database (Coltheart,1981), also see http://www.psy.uwa.edu.au/MRCDataBase/uwa_mrc.htm. These were presented in white letters on ablack background on a computer monitor placed 1 m fromparticipants (frequency range=1–7/million, length=4–9 let-ters). The stimuli subtended maximum visual angles of 5°(horizontal) and 0.6° (vertical).

The 280 words were split initially into 8 equal groups of 35words. One complete word task list comprised two study listsand two test lists. There were 70 words (2 groups) on eachstudy list. There were 175 presentations of words on each testlist. These comprised the 70 words presented at study alongwith 35 unstudied words, and a further 35 unstudied wordswhich then repeated once after 7–9 intervening words. Thenumbers of words re-presented after lags 7, 8, and 9 wereapproximately equal. For each test list there were also 6 fillersplaced towards the end of the lists. These did not appear onany study list and were included in order to ensure that thefinal repeating word fell within the 7–9 word repeat interval.No words appeared in both test lists. Rotating the groups ofwords across study and test lists so that all words werepresented at study and test as well as at test only, either asnewwords or as repeats, resulted in the creation of 8 completetask lists. The order of presentation of words in the study listswas determined randomly for each participant. In total, eachparticipant saw 502 presentations of words (140 study words,350 test words, 12 fillers).

4.2. Task procedure

The study phases in each experiment were preceded by aperiod of approximately 30 min during which participantswere fitted with an electrode cap (see below for details). Allstudy and test phases were completed in the same testingchamber. In each study phase, participants completed oneencoding task for all words. In Experiment 1, they were askedto repeat each word aloud when it appeared on the screen. Inexperiments 2 and 3, they were asked to say aloud a wordwhich rhymed with the presented study word. In all threeexperiments, an asterisk before study words initiated eachstudy trial and remained on the screen for 1000ms. The screenwas then blanked (100 ms) before the study word waspresented for 300 ms. Participants were asked to give theirresponse after the study word appeared. In Experiment 1,2500 ms intervened between the offset of the word and theappearance of the asterisk indicating the onset of the nexttrial. In experiments 2 and 3, the next trial started 1000 msafter the participant pressed any key on the response box. Forexperiments 2 and 3, participants were informed that forpolysyllabic words where a rhyme did not come immediatelyto mind (e.g. marmalade) it was acceptable to produce a wordthat rhymed only with the last syllable of the target word.

158 B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

py

Each test trial started with a fixation asterisk (500 msduration), which was removed from the screen 100ms prior topresentation of a test word (300 ms). The screen was thenblanked until the participant responded, and the next trialstarted 1000 ms after the response. In experiments 1 and 2,participants responded on one key to studied words and onanother key to new as well as repeated test words. InExperiment 3, participants responded on one key to newwords, and on another to repeated test as well as studiedwords. All test responses weremade on a key-padwith the leftand right thumbs. The thumbs used for responses werebalanced across participants and in each experiment theinstructions were to balance speed and accuracy equally.

4.3. EEG acquisition

There were some differences in the recording parametersbetween Experiment 1 and experiments 2 and 3. In allexperiments, EEG was recorded from 25 silver/silver chlorideelectrodes at midline (Fz, Cz, Pz) and left/right hemispherelocations (FP1/FP2, F7/F8, F5/F6, F3/F4, T3/T4, C5/C6, C3/C4, T5/T6, P5/P6, P3/P4, O1/O2) located according to the 10–20 system(Jasper, 1958). Additional electrodes were placed on themastoid processes. EOG was recorded from above and belowthe left eye (VEOG) and from the outer canthi (HEOG). Datawere re-referenced off-line to linked mastoids. Trials contain-ing large EOG artefact were rejected, as were trials containingA/D saturation or baseline drift exceeding ±80 μV. Other EOGblink artefacts were corrected using a linear regressionestimate (Semlitsch et al., 1986). A 7 point binomially weightedsmoothing filter was applied prior to analysis. In Experiment 1EEG (range 0.03–40 Hz; sampling rate 200 Hz) was acquiredreferenced to Fz. The data from Fz was recovered. InExperiments 2 and 3 data were acquired from 7 locations inaddition to those used in Experiment 1 (F1/F2, C1/C2, P1/P2,Oz). EEG (range 0–419 Hz; sampling rate 2048 Hz) was acquiredreferenced to linked electrodes located midway between POzand PO3/PO4, respectively. Data were high-pass filtered off-line (0.03 Hz) and down-sampled to 200 Hz. Epoched data ineach experiment comprised 200 points (1 s epoch length,including a 100 ms pre-stimulus baseline relative to which allmean amplitude values were calculated).

Appendix A. Meanacross-participant trial numberscontributing to each of the critical conditions inexperiments 1, 2 and 3 (ranges in brackets)

Exp.1 Exp.2 Exp.3

Hit 82 (45–119) 72 (26–111) 87 (66–117)Miss 35 (14–64) 35 (12–83) 32 (14–52)Correct rejection 115 (76–147) 109 (69–134) 115 (85–138)Repeated hit 48 (22–67) 47 (27–61) 55 (32–70)

R E F E R E N C E S

Azimian-Faridani, N., Wilding, E.L., 2004. An event-relatedpotential study of the revelation effect. Psychon. Bull. Rev. 11,926–931.

Azimian-Faridani, N.,Wilding, E.L., 2006. The influence of criterionshifts on electrophysiological correlates of recognitionmemory. J. Cogn. Neurosci. 18, 1075–1086.

Coltheart, M., 1981. The MRC Psycholinguistic Database. Q. J. Exp.Psychol., A 33, 497–505.

Curran, T., 2000. Brain potentials of recollection and familiarity.Mem. Cogn. 28, 923–938.

Curran, T., 2004. Effects of attention and confidence on thehypothesized ERP correlates of recollection and familiarity.Neuropsychologia 42, 1088–1106.

Curran, T., Tanaka, J.W., Weiskopf, D.M., 2002. Anelectrophysiological comparison of visual categorization andrecognition memory. Cogn. Affect. Behav. Neurosci. 2, 1–18.

Curran, T., Tepe, K.L., Piatt, C., in press. ERP explorations of dualprocesses in recognition memory. In: H.D. Zimmer, et al., Eds.,Binding in humanmemory: A neurocognitive approach. OxfordUniversity Press, Oxford.

Donaldson, D.I., Rugg, M., 1999. An event-related study comparingassociative recognition and associative recall. Cogn. Brain Res.8, 1–16.

Donchin, E., Coles, M.G.H., 1988. Is the P300 component amanifestation of context updating? Behav. Brain Sci. 11,355–372.

Dunn, J.C., 2004. Remember-know: a matter of confidence.Psychol. Rev. 111, 524–542.

Duzel, E., Yonelinas, A.P., Mangun, G.R., Heinze, H.J., Tulving, E.,1997. Event-related brain potential correlates of two states ofconscious awareness in memory. Proc. Natl. Acad. Sci. U. S. A.94, 5973–5978.

Duzel, E., Vargha-Khadem, F., Heinze, H.J., Mishkin, M., 2001. Brainactivity evidence for recognition without recollection afterearly hippocampal damage. Proc. Natl. Acad. Sci. U. S. A. 98,8101–8106.

Dywan, J., Segalowitz, S.J., Webster, L., 1998. Source monitoring:ERP evidence for greater reactivity to non-target information inolder adults. Brain Cogn. 36, 390–430.

Dywan, J., Segalowitz, S.J., Webster, L., Hendry, K., Harding, J., 2001.Event-related potential evidence for age-related differences inattentional allocation during a source monitoring task. Dev.Neuropsychol. 19, 99–120.

Dywan, J., Segalowitz, S.J., Arsenault, A., 2002. Electrophysiologicalresponse during source memory decisions in older andyounger adults. Brain Cogn. 49, 322–340.

Dzulkifli, M.A., Wilding, E.L., 2005. Electrophysiological indices ofstrategic episodic retrieval processing. Neuropsychologia 43,1152–1162.

Dzulkifli, M.A., Herron, J.E., Wilding, E.L., 2006. Memory retrievalprocessing: neural indices of processes supporting selectiveepisodic retrieval. Neuropsychologia 44, 1120–1130.

Friedman, D., Johnson, R., 2000. Event-related potential (ERP)studies of memory encoding and retrieval: a selective review.Microsc. Res. Tech. 51, 6–28.

Herron, J.E., Rugg, M.D., 2003a. Retrieval orientation and thecontrol of recollection. J. Cogn. Neurosci. 15, 843–854.

Herron, J.E., Rugg, M.D., 2003b. Strategic influences on recollectionin the exclusion task: electrophysiological evidence. Psychon.Bull. Rev. 10, 703–710.

Hintzman, D.L., Curran, T., 1994. Retrieval dynamics of recognitionmemory and frequency judgments: evidence for separateprocesses of familiarity and recall. J. Mem. Lang. 33, 1–18.

Hornberger, M., Morcom, A.M., Rugg, M.D., 2004. Neural correlatesof retrieval orientation: effects of study-test similarity. J. Cogn.Neurosci. 16, 1196–1210.

Horst, R.L., Johnson, J.R., Donchin, E., 1980. Event-related brainpotentials and subjective probability in a learning task. Mem.Cogn. 8, 476–488.

Jacoby, L.L., 1991. A process dissociation framework: separatingautomatic from intentional uses of memory. J. Mem. Lang. 30,513–541.

159B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0

Autho

r's

pers

onal

co

py

Jacoby, L.L., 1998. Invariance in automatic influences of memory:toward a user's guide for the process-dissociation procedure.J. Exper. Psychol., Learn., Mem., Cogn. 24, 3–26.

Jacoby, L.L., Kelley, C., 1992. Unconscious influences of memory:dissociations and automaticity. In: Milner, A.D., Rugg, M.D.(Eds.), The Neuropsychology of Consciousness. AcademicPress, London, pp. 201–233.

Jacoby, L.L., Woloshyn, V., Kelley, C.M., 1997. The relation betweenconscious and unconscious (automatic) influences: adeclaration of independence. J. Exp. Psychol. Gen. 118, 115–125.

Jasper, H.A., 1958. The ten–twenty system of the internationalfederation. Electroencephalogr. Clin. Neurophysiol. 10,371–375.

Jennings, J.M., Jacoby, L.L., 1997. An opposition procedure fordetecting age-related deficits in recollection: telling effects ofrepetition. Psychol. Aging 12, 352–361.

Mecklinger, A., 2000. Interfacing mind and brain: a neurocognitivemodel of recognition memory. Psychophysiology 37, 565–582.

Nessler, D., Mecklinger, A., Penney, T.B., 2001. Event related brainpotentials and illusory memories: the effects of differentialencoding. Cogn. Brain Res. 10, 283–301.

Nessler, D., Mecklinger, A., Penney, T.B., 2005. Perceptualfluency, semantic familiarity and recognition-relatedfamiliarity: an electrophysiological exploration. Cogn. BrainRes. 22, 265–288.

Neville, H.J., Kutas, M., Chesney, G., Schmidt, A.L., 1986.Event-related brain potentials during initial encoding andrecognition memory of congruous and incongruous wordsJ. Mem. Lang. 25, 75–92.

Penney, T.B., Mecklinger, A., Nessler, D., 2001. Repetition relatedERP effects in a visual object target detection task. Cogn. BrainRes. 10, 239–250.

Rotello, C.M., MacMillan, N.A., Reeder, J.A., 2004. Sum-differencetheory of remembering and knowing: a two-dimensionalsignal detection model. Psychol. Rev. 111, 588–616.

Rugg, M.D., Nagy, M.E., 1989. Event-related potentials andrecognition memory for words. Electroencephalogr. Clin.Neurophysiol. 72, 395–406.

Rugg, M.D., Roberts, R.C., Potter, D.D., Pickles, C.D., Nagy, M.E.,1991. Event-related potentials related to recognition memory.Effects of unilateral temporal lobectomy and temporal lobeepilepsy. Brain 114, 2313–2332.

Rugg, M.D., Mark, R.E., Gilchrist, J., Roberts, S.C., 1997. ERPrepetition effects in direct and indirect tasks; effects of age andinter-item lag. Psychophysiology 34, 572–586.

Rugg, M.D., Mark, R.E., Walla, P., Schloerscheidt, A., Birch, C.S.,Allan, K., 1998. Dissociation of the neural correlates of implicitand explicit memory. Nature 392, 595–598.

Rugg, M.D., Herron, J., Morcom, A., 2002. Electrophysiologicalstudies of retrieval processing. In: Squire, L.R., Schacter, D.L.(Eds.), Neuropsychology of Memory. Guilford Publications,New York, pp. 154–165.

Semlitsch, H.V., Anderer, P., Schuster, P., Presslich, O., 1986. Asolution for reliable and valid reduction of ocular artifacts,applied to the P300 ERP. Psychophysiology 23, 695–703.

Smith, M.E., 1993. Neurophysiological manifestations ofrecollective experience during recognition memoryjudgements. J. Cogn. Neurosci. 5, 1–13.

Smith, M.E., Halgren, E., 1989. Dissociation of recognition memory

components following temporal lobe lesions. J. Exper. Psychol.,Learn., Mem., Cogn. 15, 50–60.

Snodgrass, J.G., Corwin, J., 1988. Pragmatics of measuringrecognition memory: applications to dementia and amnesia.J. Exp. Psychol. Gen. 117, 34–50.

Squires, N.K., Squires, K.C., Hillyard, S.A., 1975. Two varieties oflong-latency positive waves evoked by unpredictable auditorystimuli in man. Neurophysiology 38, 387–401.

Swick, D., Knight, R.T., 1997. Event-related potentials differentiatethe effect of aging on word and nonword repetition in explicitand implicit memory tasks. J. Exper. Psychol., Learn., Mem.,Cogn. 23, 123–142.

Tendolkar, I., Schoenfeld, A., Golz, G., Fernandez, G., Kuhl, K.,Ferszt, R., Heinze, H., 1999. Neural correlates of recognitionmemory with and without recollection in patients withAlzheimer's disease and healthy controls. Neurosci. Lett. 263,45–48.

Van Petten, C., Senkfor, A.J., Newberg, W.M., 2000. Memory fordrawings in locations: spatial source memory andevent-related potentials. Psychophysiology 4, 551–564.

Wilding, E.L., 2000. In what way does the parietal ERPold/new effect index recollection? Int. J.Psychophysiol. 35, 81–87.

Wilding, E.L., Rugg, M.D., 1996. An event-related potential study ofrecognition memory with and without retrieval of source.Brain 119, 889–905.

Wilding, E.L., Rugg, M.D., 1997. Event-related potentials and therecognition memory exclusion task. Neuropsychologia 35,119–128.

Wilding, E.L., Sharpe, H., 2003. Episodic memory encoding andretrieval: recent insights from event-related potentials. In:Zani, A., Mado-Proverbio, A. (Eds.), The CognitiveElectrophysiology of Mind and Brain. Academic Press,San Diego, pp. 169–196.

Wilding, E.L., Fraser, C.S., Herron, J.E., 2005. Indexing strategicretrieval of colour information with event-related potentials.Cogn. Brain Res. 25, 19–32.

Wixted, J.T., Stretch, V., 2004. In defense of the signal-detectioninterpretation of remember/know judgments. Psychon. Bull.Rev. 11, 616–641.

Woodruff, C.C., Hayama, H.R., Rugg, M.D., 2006.Electrophysiological dissociation of the neural correlates ofrecollection and familiarity. Brain Res. 1100, 125–135.

Yonelinas, A.P., 2001a. Components of episodic memory: thecontribution of recollection and familiarity. Philos. Trans. R.Soc. London 356, 1–12.

Yonelinas, A.P., 2001b. Consciousness, control, and confidence:the 3 Cs of recognition memory. J. Exp. Psychol. Gen. 130,361–379.

Yonelinas, A.P., 2002. The nature of recollection andfamiliarity: a review of the 30 years of research. J. Mem. Lang.46, 441–517.

Yonelinas, A.P., Dobbins, I., Szymanski, M.D., Dhaliwal, H.S., King,L., 1996. Signal-detection, threshold, and dual-process modelsof recognition memory: ROCs and conscious recollection.Conscious. Cogn. 5, 418–441.

Yovel, G., Paller, K.A., 2004. The neural basis of thebutcher-on-the-bus phenomenon: when a face seemsfamiliar but is not remembered. NeuroImage 21, 789–800.

160 B R A I N R E S E A R C H 1 1 1 4 ( 2 0 0 6 ) 1 4 9 – 1 6 0