An early electrophysiological response associated …psych.colorado.edu/~tcurran/WonGauWorDebCur05.pdftivity at about 150 msec for pronounceable letter strings than for strings of

Copyright 2005 Psychonomic Society Inc 306

Cognitive Affective amp Behavioral Neuroscience2005 5 (3) 306-318

Perceptual expertise with letters is a result of pro-longed experience with print The extensive reading ex-perience taking place over the years after we become lit-erate likely modifies the way we process and perceiveindividual letters For instance as expert readers we areused to seeing print in a coherent style and can thus ex-tract font information to aid letter recognition We per-form a letter identification task better with letter stringsin the same font rather than in mixed fonts (Sanocki1987 1988) Novice readers (eg English readers view-ing Chinese characters) however are much less sensitiveto variations in font information (Gauthier Wong Hay-ward amp Cheung in press) Likewise expert readers areaccustomed to seeing letters in the context of wordsWhen we fixate on part of a word we obtain not onlyhigh-resolution information about the letters in the foveabut also low-resolution information about the parafovealletters With experience in reading we develop a strong

tendency to use the low-resolution information about theparafoveal letters to such an extent that even when high-resolution information about them is artificially madeavailable (by magnifying the parafoveal letters) we areunable to utilize this extra information (Nazir Jacobs ampOrsquoRegan 1998) Such behavioral phenomena demon-strate that our perception of letters is influenced by ourreading experience

Neural selectivity can develop as a result of perceptualexpertise with certain categories of objects (Gauthier2000) There are at least two neural hallmarks of the kindof expertise we acquire for identifying objects within homogeneous classes (eg faces cars dogs birds orcomputer-generated novel objects) In comparison withcommon objects objects with which we have expertiseelicit a larger event-related potential (ERP) componentmdashN170mdashin posterior brain regions (Rossion GauthierGoffaux Tarr amp Crommelinck 2002 Tanaka amp Curran2001) and greater recruitment of a small region in thefusiform gyrus mainly on the right (Gauthier Skud-larski Gore amp Anderson 2000 Gauthier Tarr Ander-son Skudlarski amp Gore 1999) Because the kind of ex-pertise we have with letters differs in several respectsfrom our expertise with faces cars or dogs letters shouldrecruit a different part of the extrastriate cortex Indeedas we will describe below words and letter strings (Cohenet al 2002 Polk amp Farah 1998) and more recently sin-gle letters (James James Jobard Wong amp Gauthier inpress) have been shown to elicit greater activity in partsof the left fusiform gyrus than do control stimuli in-cluding digits and unfamiliar characters

The majority of studies concerning the neural bases ofprint perception have focused on selectivity for words andpronounceable strings (Assadollahi amp Pulvermuumlller 2003

This research was supported by a grant from the James S McDonnellFoundation to the Perceptual Expertise Network NIMH Grant MH64812to TC and NEI Grant EV13441-01 to IG We are grateful to SkipJohnson for his help with interpreting the P300 effects We thank CindyBukach for her comments on an earlier version of the article two anony-mous reviewers for their constructive comments and Susan Williamsfor her assistance in experiment programming We also thank the fol-lowing lab members for help with testing subjects Sophie BoddingtonJohn Capps Tatsuko GoHollo Gabriel Matthews Christel Taylor andBrent Young An earlier version of this work was presented in a postersession at the Annual Meeting of the Visual Sciences Society at Sara-sota Florida in May 2004 Correspondence relating to this article maybe sent to A C-N Wong Department of Psychology Vanderbilt Uni-versity Nashville TN 37203 or to T Curran Department of Psychol-ogy UCB 345 University of Colorado Boulder CO 80309-0345 (e-mailalanwongvanderbiltedu or tcurranpsychcoloradoedu)

An early electrophysiological response associatedwith expertise in letter perception

ALAN C-N WONG and ISABEL GAUTHIERVanderbilt University Nashville Tennessee

and

BRION WOROCH CASEY DeBUSE and TIM CURRANUniversity of Colorado Boulder Colorado

Expertise with print is likely to optimize visual processes for recognizing characters of a familiarwriting system Although brain activations have been identified for words and letter strings in contrastwith other stimuli relatively little work has focused on the neural basis of single-letter perception En-glish readers and ChinesendashEnglish bilinguals participated in an ERP study and performed a 1-back iden-tity judgment on Roman letters Chinese characters pseudofonts and their string versions The ChinesendashEnglish bilinguals showed an enhanced N170 for both Roman letters and Chinese characters relativeto pseudofonts For the non-Chinese readers the N170 amplitude was larger for Roman letters relativeto Chinese characters and pseudofonts Our results suggest that changes in relatively early visual pro-cesses underlie expert letter perception

N170 AND LETTER EXPERTISE 307

Bookheimer 2002 Cohen et al 2000 Cohen et al2002 Dehaene Le ClecrsquoH Poline Le Bihan amp Cohen2002 Hauk amp Pulvermuumlller 2004 McCandliss Posneramp Givoacuten 1997 Petersen Fox Snyder amp Raichle 1990Proverbio Vecchi amp Zani 2004) These studies there-fore have addressed the linguistic (orthographic phono-logical or semantic) more than the perceptual aspects ofreading More relevant to the question of neural selec-tivity for letter perception per se are studies showingmore activity for unpronounceable letter strings than forcontrol stimuli For example the amplitude of the N170is greater for words pseudowords and unpronounceableconsonant strings than for strings formed by alphanu-meric symbols and forms (Bentin Mouchetant-RostaingGiard Echallier amp Pernier 1999) A greater P150 com-ponent has also been found not only for words and letterstrings but also for strings of letterlike stimuli in com-parison with object icon strings (Schendan Ganis ampKutas 1998) The P150 maximal at the central top elec-trode (Cz) when recorded with respect to a mastoid ref-erence may be the positive counterpart of the N170which is maximal at occipito-temporal electrodes Alarger intracranial N200 has also been found bilaterallyin the posterior fusiform gyrus for words and nonwords(pronounceable or not) relative to objects like cars andbutterflies (Allison McCarthy Nobre Puce amp Belger1994 Nobre Allison amp McCarthy 1994) In addition tothese electrophysiological findings fMRI has revealedmore activity in the left occipito-temporal junction forletter strings relative to textures and faces (Puce Alli-son Asgari Gore amp McCarthy 1996) Letter stringshave also been found to elicit more activity than do digitstrings with fMRI in a widespread area around the leftfusiform gyrus (Polk et al 2002) These results suggestneural selectivity for strings of letters and letterlike stim-uli that do not readily contain linguistic information at aword level although it has been argued that letter stringsare more wordlike and may evoke more word-level pro-cesses involving orthography phonology and so forththan do single letters (Price 2000)

A few other studies suggest selectivity for individualletters For example fMRI activity in bilateral occipito-temporal areas habituates to the same letter in the samefont (vs the letter in different fonts) but not to the samehuman face (vs different faces) (Gauthier Tarr et al2000) Also there is more fusiform gyrus activity forsingle letters than for oblique lines (Longcamp AntonRoth amp Velay 2003) An anterior region in the leftfusiform region has been shown to be selective for Romanletters and a more posterior region for Roman stringswith digits and Chinese characters as controls (Jameset al in press) More left middle occipital activationshave also been shown for single letters than for symbolsand colors (Flowers et al 2004 Garrett et al 2000) Aconcern is that these fMRI activations may be caused byfeedback from higher level processingmdashfor examplefrom letter naming However a number of MEG studiesconducted by Tarkiainen and colleagues have cast doubt

on this alternative account (Tarkiainen Cornelissen ampSalmelin 2002 Tarkiainen Helenius Hansen Cornelis-sen amp Salmelin 1999) These authors identified a leftinferior occipito-temporal region that showed more ac-tivity at about 150 msec for pronounceable letter stringsthan for strings of geometric shapes Despite their em-phasis on strings these studies revealed that this regionalso shows more activity for single upright letters thanfor geometric shapes The early latency of these MEGresponses makes feedback from higher level processes aless likely explanation for the selectivity found in theabove-mentioned fMRI studies

The present study examines the early neural selectivityassociated with letter expertise Two groups of subjects(English readers who cannot read Chinese and ChinesendashEnglish bilinguals) took part in an ERP experiment inwhich they saw three types of characters (Roman Chi-nese and pseudofont) The group stimulus design fea-tured expert (non-Chinese readers viewing Roman char-acters bilinguals viewing Roman or Chinese characters)and novice (non-Chinese readers viewing Chinese orpseudofont characters bilinguals viewing pseudofontcharacters) situations allowing for a more direct test ofthe association between expertise and neural selectivityfor letters For example the same stimuli (Chinese char-acters) were expected to elicit different levels of activitydepending on amount of expertisemdashthat is bilingualswere expected to show comparable activity for Romanletters and Chinese characters but non-Chinese readerswere expected to show more activity for Roman lettersthan for Chinese characters Such results would not beexplained by the feature differences between the stimuliwhich are difficult to control perfectly The use of char-acters from very different writing systems will also im-prove the generalizability of the results

We adopted ERPs as a means to tap into early visualletter processing in relative isolation from most linguis-tic processes Past research has shown that the earliestpotential to reflect high-level visual differences amongobject categories appears as a posterior negative compo-nent peaking at about 170 msec after stimulus presenta-tion (Bentin Allison Puce Perez amp McCarthy 1996Curran Tanaka amp Weiskopf 2002 Rossion Gauthieret al 2002 Tanaka amp Curran 2001) This N1N170 po-tential is associated with expertise with a visual category(Busey amp Vanderkolk 2005 Gauthier Curran Curby ampCollins 2003 Rossion Gauthier et al 2002 Tanaka ampCurran 2001) Therefore we expected a larger N170(relative to a pseudofont control) at posterior channelsfor letters with which subjects have expertisemdashthat isfor Roman letters with non-Chinese readers and for bothRoman letters and Chinese characters with bilinguals Itis important to note that the finding of an N170 effectfor letter expertise does not necessarily reflect the sameprocesses that an N170 effect found for object and faceexpertise would Since various spatiotemporally over-lapping visual processes are likely to contribute to thescalp-recorded N170 (Rossion Curran amp Gauthier

308 WONG GAUTHIER WOROCH DeBUSE AND CURRAN

2002) it is a reasonable postulate that the N170 can bemodulated by different types of perceptual expertisewith objects Our primary aim here is not to equate let-ter expertise with or dissociate it from face expertise butto describe properties of the selectivity associated withexpertise for letters and letter strings

METHOD

SubjectsThirty-seven undergraduates from the University of Colorado at

Boulder participated for course credit Twenty-two ChinesendashEnglish bilinguals participated for payment of $15h Because wewere unable to recruit as many bilingual as non-Chinese subjects thepresent results include only 18 non-Chinese readers and 18 ChinesendashEnglish bilinguals Subject selection was based upon absence ofEEG artifact (6 monolingual subjects and 1 bilingual subject wereexcluded for excessive artifact) maintaining high accuracy levelsand minimizing group differences in accuracy (subjects with lessthan 90 accuracy 6 from the non-Chinese and 3 from the ChinesendashEnglish group were excluded) maintaining counterbalancing andequating the sex distributions of the two groups (9 males and 9 fe-males per group) The ChinesendashEnglish bilinguals who were mostlygraduate students were older (mean age 24 range 19ndash29) thanthe undergraduate non-Chinese readers (mean age 19 range 18ndash22) All of the ChinesendashEnglish bilinguals were born in Chinalearned English in China (mean starting age 11 range 4ndash15)had known English for a long time (mean 13 years range 5ndash21 years) and had recently moved to the United States (mean

3 years in US range 1ndash10 years) The bilinguals were eithergraduate students or teachingresearch staff at the university Al-though we did not quantify their proficiency the fact that they allstudy or work in the United States makes it clear that they are veryfamiliar with the Roman alphabet Also the bilinguals had no prob-lems with reading consent forms communicating or comprehend-ing the experimental instructions in English

Stimuli Design and ProcedureThere were six types of stimuli (Roman Chinese and pseudofont

characters and their string versions) Figure 1 shows the eightRoman consonants eight Chinese characters and eight pseudofontcharacters used and one example of each type of trial Each char-acter measured about 1 1 cm onscreen (057ordm at a viewing dis-tance of 100 cm) Each string consisted of five characters and wasabout 7 cm wide (4ordm at a viewing distance of 100 cm) The Romanstrings were formed by randomly picking and assembling Romanletters to form 100 different five-character strings and then replac-ing characters in certain strings according to the following rules(1) No repetition of letters was permitted within a string (2) All let-ters occurred at approximately the same frequency in the 100 strings(mean 62 range 58ndash65) (3) All letters occurred at approxi-mately the same frequency (12 or 13) in the central underlined po-sition (4) No familiar or potentially meaningful two-letter combi-nations were permitted (eg HP HB BP HK) (5) No validgraphemes were permitted (eg BL PH) (6) All two-letter combi-nations (eg DF) except for the removed ones occurred at similarfrequencies (mean 796 range 6ndash12) The Chinese and pseu-dofont strings were formed by taking the 100 Roman strings and re-placing them with corresponding Chinese or pseudofont characters

Figure 1 All of the stimuli used in the experiment as well as examplesof each type of trial


We also checked to ensure that there were no meaningful charactercombinations in the Chinese strings (eg which means ldquodrysoilrdquo)

There were 100 trials for each of the six types of stimuli sepa-rated into 5 blocks containing 20 trials each The subjects per-formed a 1-back identity-matching task Each trial started with afixation cross at the center of the screen for a random period be-tween 250 and 750 msec A stimulus (a character or string) then ap-peared for 750 msec followed by a 500-msec blank screen and thefixation cross for the next trial The subjects were instructed topress ldquo1rdquo on the number keypad when the character shown wasidentical to the previous one or when the central underlined char-acter of the current string repeated that of the previous string (flank-ing characters were always different in consecutive trials) Same tri-als amounted to 10 of all trials for each stimulus (ie 10 out of100) In other nontarget trials no response was required The num-bers of same trials in the 5 blocks were 1 2 2 2 and 3 The sixtypes of stimuli resulted in a total of 600 trials presented in 30blocks Each block only contained one type of stimulus The dif-ferent stimulus blocks alternated with each other such that the sixtypes of stimulus blocks were each presented once before any ofthem was presented a second time and so forth in later blocks Theorder of block presentation was counterbalanced across subjectsForty trials (20 for Roman letters 20 for Roman strings) were in-troduced at the beginning as practice

EEGERP MethodsScalp voltages were collected with a 128-channel Geodesic Sen-

sor Net (Tucker 1993) connected to an AC-coupled 128-channelhigh-input impedance amplifier (200-MΩ Net Amps Electrical Geo-desics Inc Eugene OR) Amplified analog voltages (01ndash100 Hzbandpass 3 dB) were digitized at 250 Hz Individual sensors wereadjusted until impedances were less than 50 kΩ The EEG was dig-itally low-pass filtered at 40 Hz Trials were discarded from analy-ses if they contained incorrect responses or eye movements (EOGover 70 μV) or if more than 20 of the channels were bad (averageamplitude over 100 μV or transit amplitude over 50 μV) Target(same judgment) trials were also excluded from analyses The meannumber of trials per subject per condition was 81 (range 56ndash90)Individual bad channels were replaced on a trial-by-trial basis ac-cording to a spherical spline algorithm (Srinivasan Nunez Silber-stein Tucker amp Cadusch 1996) EEG was measured with respectto a vertex reference (Cz) but an average-reference transformationwas used to minimize the effects of reference-site activity and toestimate accurately the scalp topography of the measured electricalfields (Bertrand Perin amp Pernier 1985 Curran Tucker Kutas ampPosner 1993 Dien 1998 Lehman amp Skrandies 1985 PictonLins amp Scherg 1995 Tucker Liotti Potts Russell amp Posner1994) Average-reference ERPs were computed for each channel asthe voltage difference between that channel and the average of allchannels The average reference was corrected for the polar averagereference effect (Junghoumlfer Elbert Tucker amp Braun 1999) ERPswere baseline corrected with respect to a 100-msec prestimulusrecording interval

RESULTS

Behavioral ResultsThe two groups of subjects maintained a similarly high

level of accuracy [non-Chinese readers 96 ChinesendashEnglish bilinguals 97 t(34) 144 SE 081 p 10] Analyses of variance (ANOVAs) were conductedon both accuracy and response time data with group(non-Chinese reader or ChinesendashEnglish bilingual) stim-ulus (Chinese Roman or pseudofont) and character

string as factors When necessary in this and all subse-quently reported ANOVAs degrees of freedom were ad-justed according to the conservative GreenhousendashGeisserprocedure for sphericity violations (Winer 1971)

There was a correspondence between accuracy andlevel of expertise as revealed by a significant group stimulus interaction [F(268) 813 p 01] For non-Chinese readers accuracy was ordered from Roman(986) to pseudofont (967) to Chinese (933) stim-uli with both the Romanndashpseudofont [t(17) 449 p 01] and pseudofontndashChinese [t(17) 261 p 05] dif-ferences being significant The ChinesendashEnglish bilin-guals performed best with Chinese stimuli (983) fol-lowed by Roman (981) and pseudofont (961) stimuliand only the Romanndashpseudofont difference was signifi-cant [t(17) 236 p 03] In terms of response timeonly the group characterstring interaction was signif-icant [F(134) 778 p 05] Whereas the non-Chinesereaders responded faster for characters than for strings[485 vs 515 msec t(17) 294 p 05] no significantdifference was observed between characters and stringsfor bilinguals (482 vs 460 msec p 20)

When the data for the 11 non-Chinese readers ex-cluded on the basis of lower accuracy or the sex compo-sition of the two groups were also included similar re-sults were obtained with an additional effect of groupNot surprisingly the inclusion of the previously excludednon-Chinese readers caused the performance of non-Chinese readers to become worse than that of bilinguals[accuracy F(145) 726 p 01 response timeF(145) 397 p 052]

ERP ResultsERPs from selected 10ndash20 locations are shown in Fig-

ures 2 (non-Chinese readers viewing characters) 3 (non-Chinese readers viewing strings) 4 (ChinesendashEnglishbilinguals viewing characters) and 5 (ChinesendashEnglishbilinguals viewing strings) Overall the most outstand-ing feature is the P300 difference related to expertise(eg channel Pz between about 300 and 600 msec)P300 amplitude was smaller when subjects viewed stim-uli with which they had experience (non-Chinese readersviewing Roman stimuli bilinguals viewing Roman orChinese stimuli) than when they viewed unfamiliar stim-uli (non-Chinese readers viewing Chinese or pseudofontstimuli bilinguals viewing pseudofont stimuli) Presum-ably unfamiliar characters and strings were perceptuallymore complex and P300 amplitude is known to increasewith stimulus complexity (Johnson 1986 1993) Al-though our interest was in early visual processes formalanalyses were conducted on both the N170 and P300 effects

Subsequent analyses were conducted on ERPs aver-aged across all artifact-free nontarget trials (those towhich no response was required) For both the N170 andP300 analyses peak ERP amplitude was the primary in-dependent measure Previous studies have used meanamplitude as the primary dependent measure but peak


amplitude (ie minimum for N170 and maximum forP300) was used for the present analysis for two reasonsFirst latency differences among conditions (as will ap-pear later) may bias the results if a fixed window is usedfor calculating mean amplitude Second inspection ofthe ERPs suggested that expertise-related group stim-ulus interactions in both the N170 and P300 componentsoverlapped in time with each other so that at least insome conditions we were unable to select a N170 meanamplitude window that did not overlap with P300 andvice versa

N170 results The first step in our analysis was theidentification of the locations where the N170 compo-nent was maximal so that further analyses could focus onthese channels For each subject we computed the am-plitude of the greatest negative deflection occurring overall posterior electrode sites between 120 and 250 msecafter stimulus onset Averaged across all subjects and con-ditions the N170 was most negative for left-hemispherechannel 65 (falling between standard 10ndash20 locations T5and O1 see Figure 6) To allow for spatial variabilityacross subjects and conditions we selected a group of

Non-Chinese Readers

Time From Stimulus Onset (msec)

Roman characters

Chinese characters

Pseudofont characters

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2

ndash2 0 500 1000microV

Figure 2 ERPs from selected 10ndash20 locations for non-Chinese readers viewing characters

Non-Chinese Readers


Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2


Figure 3 ERPs from selected 10ndash20 locations for non-Chinese readers viewing strings


channels surrounding 65 for further analysis (T5 59 6465 66 70 O1) along with their right-hemisphere coun-terparts (O2 85 90 91 92 96 T6) The ERPs obtainedby averaging the channels within each region are shownin Figures 7 (non-Chinese readers) and 8 (ChinesendashEnglish bilinguals)

Minimum amplitude was entered into a group (non-Chinese readers or ChinesendashEnglish bilinguals) stim-ulus (Chinese Roman or pseudofont) characterstring hemisphere ANOVA All significant ( p 05)

results and their statistics are reported in Table 1 Over-all N170 amplitudes were more negative for the bilingualthan for the non-Chinese subjects The key result was thesignificant group stimulus interaction (Figure 9) Asignificant group characterstring interaction was alsoobserved Separate ANOVAs were conducted for the twogroups to better interpret the results

The non-Chinese subjects were considered alone in astimulus (Chinese Roman or pseudofont) charac-terstring hemisphere ANOVA Only the stimulus con-

ChinesendashEnglish Bilinguals


Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2


Figure 4 ERPs from selected 10ndash20 locations for ChinesendashEnglish bilinguals viewing characters



Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2


Figure 5 ERPs from selected 10ndash20 locations for ChinesendashEnglish bilinguals viewing strings


dition effect described previously was significant (seeTable 1) Roman stimuli led to greater N170 amplitudesthan did Chinese [t(17) 296 p 01] or pseudofont[t (17) 376 p 01] stimuli When the ChinesendashEnglish bilinguals were considered alone the stimuluscondition effect was again significant Whereas Romanand Chinese stimuli did not differ from each other (t 1)

they both led to larger N170 amplitudes than did pseudo-fonts [Roman vs pseudofonts t (17) 225 p 05Chinese vs pseudofonts t(17) 388 p 001] Thestimulus hemisphere interaction suggested that thesecondition effects were more pronounced over the left hemi-sphere Also N170s were significantly more negative forstrings than for characters

Other significant effects for both groups include themain effects of stimulus and characterstring and thestimulus hemisphere and stimulus characterstring hemisphere interactions The interactions werecaused by N170s being more negative for Roman andChinese than for pseudofont stimuli in the left hemi-sphere whereas in the right hemisphere such differencesonly occurred for characters but for not strings (all ps 05 for significant differences)

The latency of the minimum N170 was entered into a group stimulus characterstring hemisphereANOVA (Table 2) The N170 was faster for the left(176 msec) than for the right (185 msec) hemisphere andfor strings (178 msec) than for characters (183 msec)and these factors interacted in such a way that the latencydifference between characters and strings was only sig-nificant for the left hemisphere A significant group characterstring interaction was also found Whereascharacters and strings did not differ from each other fornon-Chinese readers (t 1) longer latencies were observed for characters (164 msec) than for strings(152 msec) among bilinguals [t (17) 332 p 01]The group stimulus hemisphere interaction wasalso signif icant Subsequent comparisons however

Figure 6 Channels selected for analyses (black)

Non-Chinese Readers

Roman Chinese Pseudofont

Left

Left

Characters Characters

Strings Strings

Right

Right

0 500 1000


3

2

1

0

ndash1

ndash2

ndash3

microV

Figure 7 ERPs from averaging the selected channels for non-Chinese readers


showed only a significant difference between Romanand pseudofont stimuli in the right hemisphere for bilin-guals [182 vs 188 msec t(17) 273 p 05] Acrossthe 24 conditions the mean latencies ranged from168ndash191 msec

Results were similar when the data from the 11 previ-ously excluded non-Chinese readers were included Alleffects reported above remained significant except forthe effect of characterstring in N170 latency

P300 results The P300 analyses focused on channelsFz Cz and Pz Figures 2ndash5 show the ERPs for thesechannels Maximum amplitude was entered into a group(non-Chinese readers or ChinesendashEnglish bilinguals) stimulus (Chinese Roman or pseudofont) charac-

terstring channel (Fz Cz or Pz) ANOVA (Table 3)There were significant interactions between group andstimulus and between group and channel There werealso significant group stimulus channel and group characterstring channel interactions Separate analy-ses were thus performed for each group

The non-Chinese subjects were considered alone in astimulus characterstring channel ANOVA (seeTable 3) There was a stimulus channel interactionP300 amplitude was smaller for Roman than for Chineseand pseudofont stimuli in both the Cz and Fz channels[all ts(17) 500 ps 01] The characterstring channel interaction was also significant P300 amplitudewas smaller for characters than for strings at Pz [t(17)


Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV

Figure 8 ERPs from averaging the selected channels for ChinesendashEnglish bilinguals

Table 1ANOVA Results on N170 Amplitude

Effect df F MSe p

Both Groups

Group 134 1011 1427 01Stimulus 268 998 127 001Group stimulus 268 593 127 01Characterstring (CRST) 134 1950 286 001Group CRST 134 481 286 05Stimulus hemisphere 268 473 038 05Stimulus CRST hemisphere 268 428 019 05

Non-Chinese Readers

Stimulus 234 957 151 01


Stimulus 234 560 103 01CRST 117 2225 281 001Stimulus hemisphere 234 571 032 01


263 p 05] but higher for characters than for stringsat Fz [t(17) 341 p 01] When the ChinesendashEnglishbilinguals were considered alone an effect of stimulusemerged with P300 amplitudes being smaller for Chi-nese than for pseudofont stimuli [t(17) 429 p 05]and for Roman than for pseudofont stimuli [t (17) 458 p 05] The effect of characterstring was alsosignificant with greater amplitude for characters thanfor strings

P300 peak latency was also entered into a group stimulus characterstring channel ANOVA (Table 4)The effect of stimulus was significant Latencies forRoman stimuli were shorter than those for Chinese[t(17) 423 p 05] and for pseudofont [t(17) 291p 05] stimuli The effect of characterstring was alsosignificant with characters having shorter latencies thanstrings did There was also a group channel inter-action with latencies at Pz being shorter than those atCz for non-Chinese readers and latencies at Fz beingshorter than those at Cz for bilinguals

Summary of main results The effects of primary in-terest consist of the differences in N170 amplitude among

stimuli as a function of expertise levelmdashthat is thegroup stimulus interaction Figure 10 shows topo-graphic distributions of N170 amplitude differences be-tween Roman Chinese and pseudofont stimuli at 168and 192 msec the shortest and longest N170 peak laten-cies across all conditions (and rounded to the nearest4-msec time samples) In keeping with the N170 analy-ses more negative amplitudes were observed at 168 msecin situations involving expertise (non-Chinese readersviewing Roman stimuli bilinguals viewing Roman andChinese stimuli) at posterior and inferior channels Theseexpertise effects are especially evident over the left hemi-sphere although the hemisphere stimulus interactiononly reached significance for bilinguals The differencesamong stimuli have started to extend to the superiorchannels for non-Chinese readers at 192 msec

Both the N170 and P300 peak amplitudes revealed anexpertise effect We are mainly interested in the earlierN170 effect and a potential concern is that a P300 effectwith an early onset could cause early differences amongstimuli that would be mistaken for an N170 effect Thisis unlikely however for two reasons First in the ERPplots for the channels selected for N170 analyses (Fig-ures 7 and 8) the curves for different stimuli convergedbetween the N170 and P300 peaks If P300 were drivingthe effects we would expect the differences among stim-uli to grow larger from 150 msec onward Second cor-relation analyses across subjects were conducted be-tween the N170 (left or right hemisphere) and P300 (FzCz or Pz) peak latencies across the six stimulus types(Chinese Roman and pseudofont characters or strings)Only 4 of the 36 correlations were significant ( p 05uncorrected) suggesting that the N170 and P300 effectswere separate from each other

Another result worth mentioning is the group charac-terstring interaction Whereas an N170 with a larger am-plitude and shorter latency was found with strings thanwith characters for bilinguals there was no such differencefor non-Chinese readers Interestingly this interaction didnot depend on the type of stimulus presented (Roman Chi-nese or pseudofont) Although Tarkiainen and colleagues(Tarkiainen et al 1999) obtained greater early MEG ac-tivity as the number of letters in a string increased thisdoes not explain why in our experiment stringcharacterdifferences only occurred for bilinguals Past studies havesuggested that bilingualism causes changes in ortho-graphic processing such as an increase in lateralization

English ChinesendashEnglish

Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV

Figure 9 Averages of minimum N170 amplitudes Error barsshow standard errors of the Chinesendashpseudofont and Romanndashpseudofont differences

Table 2ANOVA Results on N170 Latency for Both Groups

Effect df F MSe p

Characterstring (CRST) 134 776 41968 01Group CRST 134 469 41968 05Hemisphere 134 1560 53897 001Hemisphere CRST 134 717 9428 05Group stimulus hemisphere 268 433 8664 05


(Ding et al 2003 Hoosain 1992) To understand whetherexperience with a second language and with Chinese inparticular may lead to general changes in visual process-ing will require further experimentation

DISCUSSION

To our knowledge this is the first study showing se-lectivity of the N170 component to individual letters as-sociated with expertise The present results are consis-tent with and complement previous findings in severalways First a larger N170 was shown with Roman lettersthan with pseudofont characters for all subjects Thisearly component selective for individual letters and let-ter strings suggests that the selectivity found in otherfMRI studies (Flowers et al 2004 James et al in pressLongcamp et al 2003) was not solely caused by feed-back from higher level areas related to linguistic pro-cessing or letter name knowledge Second the group stimulus design of our study bypassed the problem ofchoosing a well-designed control stimulus The sameChinese characters resulted in either a smaller N170 am-plitude than with Roman letters (in non-Chinese readers)or a comparable amplitude (in ChinesendashEnglish bilin-guals) depending on whether subjects were experiencedwith the Chinese characters or not This expertise-associated letter selectivity cannot be explained by stim-ulus differences Third the use of Chinese characters en-

ables us to generalize our results to characters in a verydifferent writing system Fourth the present results withletters demonstrate expertise effects on the N170 that aresimilar to expertise effects previously demonstrated withobjects (Gauthier et al 2003 Rossion Gauthier et al2002 Tanaka amp Curran 2001) Finally the stronger ex-pertise effect in the left hemisphere as suggested by thestimulus hemisphere interaction on N170 amplitudein the ChinesendashEnglish bilinguals and the topographicdistributions (Figure 10) is consistent with Tarkiainenand colleaguesrsquo (Tarkiainen et al 2002 Tarkiainen et al1999) finding of a left-hemisphere preponderance forletter and letter string selectivity

Letter Expertise and Linguistic EffectsThe neural selectivity that we found for individual let-

ters is unlikely to reflect in a direct fashion language-related processes such as those invoked by words andpotentially also by nonword strings ERP componentsother than the N170 are generally found to be sensitiveto information at the word level such as orthography(Proverbio et al 2004) phonology (Bentin et al 1999)and semantics (Bentin et al 1999 McLaughlin Oster-hout amp Kim 2004) The early N170 selectivity for indi-vidual letters is likely to be free from these linguistic ef-fects because (1) these linguistic factors typically havea late effect occurring after 300 msec (except for the lex-ical frequency and orthography effects discussed below)and (2) we have shown selectivity with single Roman let-ters which do not contain the linguistic information in-volved with a word or multiple letters There does re-main a possibility that the selectivity we found at leastfor Chinese characters reflects linguistic processing at acharacter level For example Perfetti and colleagues(eg Perfetti amp Tan 1998) have found that orthographicand phonological processing started with individual Chi-

Table 4ANOVA Results on P300 Latency for Both Groups

Effect df F MSe p

Stimulus 268 848 528566 001Characterstring 134 847 597880 01Channel 268 416 836262 05Group channel 268 1147 836262 001

Table 3ANOVA Results on P300 Amplitude

Effect df F MSe p

Both Groups

Stimulus 268 3739 278 001Group stimulus 268 1617 278 001Channel 268 10393 795 001Group channel 268 3604 795 001Stimulus channel 4136 1883 071 001Group stimulus channel 4136 801 071 001Channel characterstring (CRST) 268 705 064 01Group channel CRST 268 1132 064 001

Non-Chinese Readers

Stimulus 234 3499 320 001Channel 234 10154 1025 001Stimulus channel 468 2226 081 001CRST channel 234 1270 087 001


Stimulus 234 1567 237 001CRST 117 195 569 05Channel 234 1280 566 001


nese characters within 100 msec after stimulus presenta-tion However in their recent ERP study (Liu amp Perfetti2003) the phonological processing component found forChinese characters did not appear until 400 msec afterstimulus onset

It is worth mentioning that some linguistic factorssuch as lexical frequency and orthography do have aneffect on early ERP components (Hauk amp Pulvermuumlller2004 McCandliss et al 1997 Proverbio et al 2004Sereno Brewer amp OrsquoDonnell 2003 Sereno Rayner ampPosner 1998) The effect of lexical frequency is partic-ularly interesting since it apparently contradicts our ex-pertise effect Studies have found a larger N1N170 forlow- than for high-frequency words (Hauk amp Pulver-muumlller 2004 McCandliss et al 1997 Proverbio et al2004 Sereno et al 2003 Sereno et al 1998) In con-trast our results show a larger N170 for more familiarexpert characters than for nonexpert characters It hasbeen suggested that the greater difficulty of processinglow- rather than high-frequency words may lead to alarger N1 for the low-frequency words (Sereno et al2003) In that case words of different frequencies utilizethe same neural substrates The situation may be differ-ent in this study in that expert and nonexpert charactersmay be treated as different types of stimuli potentiallyby partly dissociable neuronal ensembles The largerN170 for characters with which subjects have expertisemay indicate that additional substrates are recruited forthem

Letter Expertise and Object ExpertisePast studies have shown a greater N170 for faces than

for other common objects (Bentin et al 1996 Rossionet al 2000) Enhanced N170 components have also beenobserved when people develop expertise with cars dogsbirds and even novel objects (Curran et al 2002 Ros-sion Gauthier et al 2002 Tanaka amp Curran 2001) Thepresent results generalize this effect to another stimulusdomainmdashindividual letters One question is whether thesame processes underlie expertise with letters and withother categories of objects The process map hypothesisof cortical organization provides a framework for con-sidering this question (Gauthier 2000) According tothis hypothesis one develops neural selectivity to an ob-ject class after prolonged experience of processing theobjects in a specific manner In other words the specificneural selectivity is related to the specific constraints ofthe task associated with the category Following thislogic letter expertise may be regarded as different fromobject expertise because different computational de-mands are involved In order to perceive letters duringreading readers need to perceive a particular letter asfor instance a g and not an f irrespective of any physi-cal differences like font size color and so on For ex-pertise with faces and many other categories of objectshowever people usually gain experience in discriminat-ing among very similar objects of a homogeneous class(eg telling one face from another or distinguishing be-tween two different bird species) and it is thought that

168 msec 192 msecNon-Chinese Bilingual Non-Chinese Bilingual

RomanndashPseudofont

ChinesendashPseudofont

RomanndashChinese

ndash1 microV +1 microV

Figure 10 Topographic distributions of ERP differences between stimuli for non-Chinese read-ers and ChinesendashEnglish bilingual readers at 168 and 192 msec which represent the range of N170latencies for different conditions


the resulting expertise relies on holistic and configuralprocesses (Diamond amp Carey 1986) The N170 compo-nent has recently been associated with holistic process-ing in car experts (Gauthier et al 2003) and fingerprintexperts (Busey amp Vanderkolk 2005) The differences intask demands between objects and letters suggest thatdifferent neural and behavioral phenomena may be foundfor these two types of expertise Indeed the two types ofexpertise seem to be supported by different neural sub-strates as shown in an fMRI study (Gauthier Tarr et al2000)

Although the above question is worth examining itshould be noted that distinguishing expertise for lettersfrom the expertise we have with faces or other objectswas not the purpose of this study especially given thelimited spatial resolution of the ERP technique In factwe did not f ind any signif icant differences between experience-associated selectivity for individual lettersand letter strings in terms of the topography of their ac-tivations despite the different loci of letter- and string-selective regions shown in recent fMRI results (Jameset al in press) We can only conclude that the visual pro-cessing associated with expertise for letters letter stringsfaces and other objects as well (eg birds dogs or cars)appears to occur within the same time window

Up to now single-letter recognition has not been thefocus of reading or object recognition studies but our re-sults suggest that it may be an important avenue for fu-ture exploration Psychophysical studies have shown thatperformance in word recognition depends on how wellindividual letters are identified (Nazir et al 1998 PelliFarell amp Moore 2003) Some studies of pure alexia havelinked that reading disorder to deficits in letter recogni-tion (Arguin Fiset amp Bub 2002 Saffran amp Coslett1998) Accordingly understanding letter recognition isone important step in understanding the reading processIn addition the perception of letters distinguishes itselffrom that of other shapes and objects as indicated bysome unique behavioral phenomena (Gauthier et al inpress Sanocki 1987 1988) The present study providesanother source of support for the existence of special-ization for individual letter perception in the visual sys-tem by providing evidence of early neural selectivity forfamiliar individual letters In the future efforts to com-pare processes and neural substrates supporting the per-ception of common objects and objects of different typesof expertise (eg letters or faces) should further our un-derstanding of visual object recognition and how itchanges when we acquire expertise in various domains

REFERENCES

Allison T McCarthy G Nobre A Puce A amp Belger A(1994) Human extrastriate visual cortex and the perception of faceswords numbers and colors Cerebral Cortex 4 544-554

Arguin M Fiset S amp Bub D (2002) Sequential and parallel letterprocessing in letter-by-letter dyslexia Cognitive Neuropsychology19 535-555

Assadollahi R amp Pulvermuumlller F (2003) Early influences of

word length and frequency A group study using MEG NeuroReport14 1183-1187

Bentin S Allison T Puce A Perez E amp McCarthy G (1996)Electrophysiological studies of face perception in humans Journal ofCognitive Neuroscience 8 551-565

Bentin S Mouchetant-Rostaing Y Giard M H EchallierJ F amp Pernier J (1999) ERP manifestations of processing printedwords at different psycholinguistic levels Time course and scalp dis-tribution Journal of Cognitive Neuroscience 11 235-260

Bertrand O Perin F amp Pernier J (1985) A theoretical justifica-tion of the average reference in topographic evoked potential studiesElectroencephalography amp Clinical Neuroscience 62 462-464

Bookheimer S Y (2002) Functional MRI of language New ap-proaches to understanding the cortical organization of semantic pro-cessing Annual Review of Neuroscience 25 151-188

Busey T A amp Vanderkolk J R (2005) Behavioral and electro-physiological evidence for configural processing in fingerprint ex-perts Vision Research 45 431-448

Cohen L Dehaene S Naccache L Leheacutericy S Dehaene-Lambertz G Henaff M A amp Michel F (2000) The visualword form area Spatial and temporal characterization of an initialstage of reading in normal subjects and posterior split-brain patientsBrain 123 291-307

Cohen L Leheacutericy S Chochon F Lemer C Rivaud S amp De-haene S (2002) Language-specific tuning of visual cortex Func-tional properties of the visual word form area Brain 125 1054-1069

Curran T Tanaka J W amp Weiskopf D M (2002) An electro-physiological comparison of visual categorization and recognitionmemory Cognitive Affective amp Behavioral Neuroscience 2 1-18

Curran T Tucker D M Kutas M amp Posner M I (1993)Topography of the N400 Brain electrical activity reflecting seman-tic expectancy Electroencephalography amp Clinical Neurophysiology88 188-209

Dehaene S Le ClecrsquoH G Poline J B Le Bihan D amp Cohen L(2002) The visual word form area A prelexical representation of vi-sual words in the fusiform gyrus NeuroReport 13 321-325

Diamond R amp Carey S (1986) Why faces are and are not specialAn effect of expertise Journal of Experimental Psychology General115 107-117

Dien J (1998) Issues in the application of the average reference Re-view critiques and recommendations Behavior Research MethodsInstruments amp Computers 30 34-43

Ding G Perry C Peng D Ma L Li D Xu S Luo Q Xu Damp Yang J (2003) Neural mechanisms underlying semantic and or-thographic processing in ChinesendashEnglish bilinguals NeuroReport14 1557-1562

Flowers D L Jones K Noble K VanMeter J Zeffiro T AWood F B amp Eden G F (2004) Attention to single letters acti-vates left extrastriate cortex NeuroImage 21 829-839

Garrett A S Flowers D L Absher J R Fahey F H GageH D Keyes J W Porrino L J amp Wood F B (2000) Corticalactivity related to accuracy of letter recognition NeuroImage 11111-123

Gauthier I (2000) What constrains the organization of the ventraltemporal cortex Trends in Cognitive Sciences 4 1-2

Gauthier I Curran T Curby K M amp Collins D (2003) Per-ceptual interference supports a non-modular account of face pro-cessing Nature Neuroscience 6 428-432

Gauthier I Skudlarski P Gore J C amp Anderson A W (2000)Expertise for cars and birds recruits brain areas involved in facerecognition Nature Neuroscience 3 191-197

Gauthier I Tarr M J Anderson A W Skudlarski P amp GoreJ C (1999) Activation of the middle fusiform ldquoface areardquo increaseswith expertise in recognizing novel objects Nature Neuroscience 2568-573

Gauthier I Tarr M J Moylan J Anderson A W Skud-larski P amp Gore J C (2000) The fusiform ldquoface areardquo is part ofa network that processes faces at the individual level Journal of Cog-nitive Neuroscience 12 495-504

Gauthier I Wong A C-N Hayward W G amp Cheung O S-C


(in press) Font-tuning associated with expertise in letter perceptionPerception

Hauk O amp Pulvermuumlller F (2004) Effects of word length and fre-quency on the human event-related potential Clinical Neurophysiol-ogy 115 1090-1103

Hoosain R (1992) Differential cerebral lateralization of ChinesendashEnglish bilingual functions In R J Harris (Ed) Cognitive process-ing in bilinguals (pp 561-571) Amsterdam Elsevier North-Holland

James K H James T W Jobard G Wong A C-N amp Gau-thier I (in press) Letter processing in the visual system Differentactivation patterns for single letters and strings Cognitive Affectiveamp Behavioral Neuroscience

Johnson R Jr (1986) A triarchic model of P300 amplitude Psy-chophysiology 23 367-384

Johnson R Jr (1993) On the neural generators of the P300 compo-nent of the event-related potential Psychophysiology 30 90-97

Junghoumlfer M Elbert T Tucker D M amp Braun C (1999) Thepolar average reference effect A bias in estimating the head surfaceintegral in EEG recording Clinical Neurophysiology 110 1149-1155

Lehman D amp Skrandies W (1985) Spatial analysis of evoked po-tentials in manmdashA review Progress in Neurobiology 23 227-250

Liu Y amp Perfetti C A (2003) The time course of brain activity inreading English and Chinese An ERP study of Chinese bilingualsHuman Brain Mapping 18 167-175

Longcamp M Anton J-L Roth M amp Velay J-L (2003) Visualpresentation of single letters activates a premotor area involved inwriting NeuroImage 19 1492-1500

McCandliss B D Posner M I amp Givoacuten T (1997) Brain plastic-ity in learning visual words Cognitive Psychology 33 88-110

McLaughlin J Osterhout L amp Kim A (2004) Neural correlatesof second-language word learning Minimal instruction producesrapid change Nature Neuroscience 7 703-704

Nazir T A Jacobs A M amp OrsquoRegan J K (1998) Letter legibil-ity and visual word recognition Memory amp Cognition 26 810-821

Nobre A C Allison T amp McCarthy G (1994) Word recognitionin the human inferior temporal lobe Nature 372 260-263

Pelli D G Farell B amp Moore D C (2003) The remarkable in-efficiency of word recognition Nature 423 752-756

Perfetti C A amp Tan L H (1998) The time course of graphicphonological and semantic activation in Chinese character identifi-cation Journal of Experimental Psychology Learning Memory ampCognition 24 101-118

Petersen S E Fox P T Snyder A Z amp Raichle M E (1990)Activation of extrastriate and frontal cortical areas by visual wordsand word-like stimuli Science 249 1041-1044

Picton T W Lins O G amp Scherg M (1995) The recording andanalysis of event-related potentials In J Grafman (Ed) Handbook ofneuropsychology (Vol 10 pp 3-73) Amsterdam Elsevier

Polk T A amp Farah M J (1998) The neural development and orga-nization of letter recognition Evidence from functional neuroimagingcomputational modeling and behavioral studies Proceedings of theNational Academy of Sciences 95 847-852

Polk T A Stallcup M Aguirre G K Alsop D C DrsquoEspo-sito M Detre J A amp Farah M J (2002) Neural specializationfor letter recognition Journal of Cognitive Neuroscience 14 145-159

Price C J (2000) The anatomy of language Contributions from func-tional neuroimaging Journal of Anatomy 197 335-359

Proverbio A M Vecchi L amp Zani A (2004) From orthography tophonetics ERP measures of grapheme-to-phoneme conversion mech-anisms in reading Journal of Cognitive Neuroscience 16 301-317

Puce A Allison T Asgari M Gore J C amp McCarthy G(1996) Differential sensitivity of human visual cortex to faces let-terstrings and textures A functional magnetic resonance imagingstudy Journal of Neuroscience 16 5205-5215

Rossion B Curran T amp Gauthier I (2002) A defense of the subordinate-level expertise account for the N170 component Cogni-tion 85 189-196

Rossion B Gauthier I Goffaux V Tarr M J amp Cromme-linck M (2002) Expertise training with novel objects leads to left-lateralized facelike electrophysiological responses PsychologicalScience 13 250-257

Rossion B Gauthier I Tarr M J Despland P Bruyer RLinotte S amp Crommelinck M (2000) The N170 occipito-temporal component is delayed and enhanced to inverted faces butnot to inverted objects An electrophysiological account of face-specific processes in the human brain NeuroReport 11 69-74

Saffran E M amp Coslett H B (1998) Implicit vs letter-by-letterreading in pure alexia A tale of two systems Cognitive Neuro-psychology 15 141-166

Sanocki T (1987) Visual knowledge underlying letter perceptionFont-specific schematic tuning Journal of Experimental Psychol-ogy Human Perception amp Performance 13 267-278

Sanocki T (1988) Font regularity constraints on the process of letterrecognition Journal of Experimental Psychology Human Perceptionamp Performance 14 472-480

Schendan H E Ganis G amp Kutas M (1998) Neurophysiologicalevidence for visual perceptual categorization of words and faceswithin 150 msec Psychophysiology 35 240-251

Sereno S C Brewer C C amp OrsquoDonnell P J (2003) Context ef-fects in word recognition Evidence for early interactive processingPsychological Science 14 328-333

Sereno S C Rayner K amp Posner M (1998) Establishing a time-line of word recognition Evidence from eye movements and event-related potentials NeuroReport 9 2195-2200

Srinivasan R Nunez P L Silberstein R B Tucker D M ampCadusch P J (1996) Spatial sampling and filtering of EEG withspline-Laplacians to estimate cortical potentials Brain Topography8 355-366

Tanaka J W amp Curran T (2001) A neural basis for expert objectrecognition Psychological Science 12 43-47

Tarkiainen A Cornelissen P L amp Salmelin R (2002) Dynam-ics of visual feature analysis and object-level processing in face ver-sus letter-string perception Brain 125 1125-1136

Tarkiainen A Helenius P Hansen P C Cornelissen P L ampSalmelin R (1999) Dynamics of letter string perception in thehuman occipitotemporal cortex Brain 122 2119-2132

Tucker D M (1993) Spatial sampling of head electrical fields Thegeodesic sensor net Electroencephalography amp Clinical Neurophys-iology 87 154-163

Tucker D M Liotti M Potts G F Russell G S amp PosnerM I (1994) Spatiotemporal analysis of brain electrical fields HumanBrain Mapping 1 134-152

Winer B J (1971) Statistical principles in experimental design (2nded) New York McGraw-Hill

(Manuscript received September 1 2004revision accepted for publication May 9 2005)


Bookheimer 2002 Cohen et al 2000 Cohen et al2002 Dehaene Le ClecrsquoH Poline Le Bihan amp Cohen2002 Hauk amp Pulvermuumlller 2004 McCandliss Posneramp Givoacuten 1997 Petersen Fox Snyder amp Raichle 1990Proverbio Vecchi amp Zani 2004) These studies there-fore have addressed the linguistic (orthographic phono-logical or semantic) more than the perceptual aspects ofreading More relevant to the question of neural selec-tivity for letter perception per se are studies showingmore activity for unpronounceable letter strings than forcontrol stimuli For example the amplitude of the N170is greater for words pseudowords and unpronounceableconsonant strings than for strings formed by alphanu-meric symbols and forms (Bentin Mouchetant-RostaingGiard Echallier amp Pernier 1999) A greater P150 com-ponent has also been found not only for words and letterstrings but also for strings of letterlike stimuli in com-parison with object icon strings (Schendan Ganis ampKutas 1998) The P150 maximal at the central top elec-trode (Cz) when recorded with respect to a mastoid ref-erence may be the positive counterpart of the N170which is maximal at occipito-temporal electrodes Alarger intracranial N200 has also been found bilaterallyin the posterior fusiform gyrus for words and nonwords(pronounceable or not) relative to objects like cars andbutterflies (Allison McCarthy Nobre Puce amp Belger1994 Nobre Allison amp McCarthy 1994) In addition tothese electrophysiological findings fMRI has revealedmore activity in the left occipito-temporal junction forletter strings relative to textures and faces (Puce Alli-son Asgari Gore amp McCarthy 1996) Letter stringshave also been found to elicit more activity than do digitstrings with fMRI in a widespread area around the leftfusiform gyrus (Polk et al 2002) These results suggestneural selectivity for strings of letters and letterlike stim-uli that do not readily contain linguistic information at aword level although it has been argued that letter stringsare more wordlike and may evoke more word-level pro-cesses involving orthography phonology and so forththan do single letters (Price 2000)

A few other studies suggest selectivity for individualletters For example fMRI activity in bilateral occipito-temporal areas habituates to the same letter in the samefont (vs the letter in different fonts) but not to the samehuman face (vs different faces) (Gauthier Tarr et al2000) Also there is more fusiform gyrus activity forsingle letters than for oblique lines (Longcamp AntonRoth amp Velay 2003) An anterior region in the leftfusiform region has been shown to be selective for Romanletters and a more posterior region for Roman stringswith digits and Chinese characters as controls (Jameset al in press) More left middle occipital activationshave also been shown for single letters than for symbolsand colors (Flowers et al 2004 Garrett et al 2000) Aconcern is that these fMRI activations may be caused byfeedback from higher level processingmdashfor examplefrom letter naming However a number of MEG studiesconducted by Tarkiainen and colleagues have cast doubt

on this alternative account (Tarkiainen Cornelissen ampSalmelin 2002 Tarkiainen Helenius Hansen Cornelis-sen amp Salmelin 1999) These authors identified a leftinferior occipito-temporal region that showed more ac-tivity at about 150 msec for pronounceable letter stringsthan for strings of geometric shapes Despite their em-phasis on strings these studies revealed that this regionalso shows more activity for single upright letters thanfor geometric shapes The early latency of these MEGresponses makes feedback from higher level processes aless likely explanation for the selectivity found in theabove-mentioned fMRI studies

The present study examines the early neural selectivityassociated with letter expertise Two groups of subjects(English readers who cannot read Chinese and ChinesendashEnglish bilinguals) took part in an ERP experiment inwhich they saw three types of characters (Roman Chi-nese and pseudofont) The group stimulus design fea-tured expert (non-Chinese readers viewing Roman char-acters bilinguals viewing Roman or Chinese characters)and novice (non-Chinese readers viewing Chinese orpseudofont characters bilinguals viewing pseudofontcharacters) situations allowing for a more direct test ofthe association between expertise and neural selectivityfor letters For example the same stimuli (Chinese char-acters) were expected to elicit different levels of activitydepending on amount of expertisemdashthat is bilingualswere expected to show comparable activity for Romanletters and Chinese characters but non-Chinese readerswere expected to show more activity for Roman lettersthan for Chinese characters Such results would not beexplained by the feature differences between the stimuliwhich are difficult to control perfectly The use of char-acters from very different writing systems will also im-prove the generalizability of the results

We adopted ERPs as a means to tap into early visualletter processing in relative isolation from most linguis-tic processes Past research has shown that the earliestpotential to reflect high-level visual differences amongobject categories appears as a posterior negative compo-nent peaking at about 170 msec after stimulus presenta-tion (Bentin Allison Puce Perez amp McCarthy 1996Curran Tanaka amp Weiskopf 2002 Rossion Gauthieret al 2002 Tanaka amp Curran 2001) This N1N170 po-tential is associated with expertise with a visual category(Busey amp Vanderkolk 2005 Gauthier Curran Curby ampCollins 2003 Rossion Gauthier et al 2002 Tanaka ampCurran 2001) Therefore we expected a larger N170(relative to a pseudofont control) at posterior channelsfor letters with which subjects have expertisemdashthat isfor Roman letters with non-Chinese readers and for bothRoman letters and Chinese characters with bilinguals Itis important to note that the finding of an N170 effectfor letter expertise does not necessarily reflect the sameprocesses that an N170 effect found for object and faceexpertise would Since various spatiotemporally over-lapping visual processes are likely to contribute to thescalp-recorded N170 (Rossion Curran amp Gauthier



METHOD












RESULTS












Non-Chinese Readers


Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2



Non-Chinese Readers


Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2










Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2





Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2









Non-Chinese Readers


Left

Left


Strings Strings

Right

Right

0 500 1000


3

2

1

0

ndash1

ndash2

ndash3

microV









Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV



Effect df F MSe p

Both Groups


Non-Chinese Readers

Stimulus 234 957 151 01











Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES





































































METHOD












RESULTS












Non-Chinese Readers


Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2



Non-Chinese Readers


Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2










Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2





Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2









Non-Chinese Readers


Left

Left


Strings Strings

Right

Right

0 500 1000


3

2

1

0

ndash1

ndash2

ndash3

microV









Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV



Effect df F MSe p

Both Groups


Non-Chinese Readers

Stimulus 234 957 151 01











Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES








































































RESULTS












Non-Chinese Readers


Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2



Non-Chinese Readers


Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2










Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2





Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2









Non-Chinese Readers


Left

Left


Strings Strings

Right

Right

0 500 1000


3

2

1

0

ndash1

ndash2

ndash3

microV









Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV



Effect df F MSe p

Both Groups


Non-Chinese Readers

Stimulus 234 957 151 01











Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES






































































Non-Chinese Readers


Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2



Non-Chinese Readers


Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2










Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2





Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2









Non-Chinese Readers


Left

Left


Strings Strings

Right

Right

0 500 1000


3

2

1

0

ndash1

ndash2

ndash3

microV









Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV



Effect df F MSe p

Both Groups


Non-Chinese Readers

Stimulus 234 957 151 01











Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES










































































Roman characters

Chinese characters


Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2





Roman strings

Chinese strings

Pseudofont strings

Fp1

F7

T3

P3

T5

O1

Fz Fp2

F8

T4Cz

Pz P4

T6

O2

2









Non-Chinese Readers


Left

Left


Strings Strings

Right

Right

0 500 1000


3

2

1

0

ndash1

ndash2

ndash3

microV









Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV



Effect df F MSe p

Both Groups


Non-Chinese Readers

Stimulus 234 957 151 01











Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES









































































Non-Chinese Readers


Left

Left


Strings Strings

Right

Right

0 500 1000


3

2

1

0

ndash1

ndash2

ndash3

microV









Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV



Effect df F MSe p

Both Groups


Non-Chinese Readers

Stimulus 234 957 151 01











Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES










































































Roman

Chinese

Pseudofont

Left

Left


Strings Strings

Right

Right

0 500 1000


2

1

0

ndash1

ndash2

ndash3

ndash4

microV



Effect df F MSe p

Both Groups


Non-Chinese Readers

Stimulus 234 957 151 01











Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES











































































Roman

Chinese

Pseudofont

0

ndash1

ndash2

ndash3

ndash4

ndash5

ndash6

microV



Effect df F MSe p




DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES





































































DISCUSSION






Effect df F MSe p



Effect df F MSe p

Both Groups


Non-Chinese Readers












RomanndashChinese







REFERENCES











































































RomanndashChinese







REFERENCES







































































REFERENCES












































































































Documents

An early electrophysiological response associated …psych.colorado.edu/~tcurran/WonGauWorDebCur05.pdftivity at about 150 msec for pronounceable letter strings than for strings of