B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

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

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

Research Report

ERP evidence for the split fovea theory

Clara D. Martina,b,c,⁎, Guillaume Thierryb,d, Jean-François Démonet c,Mark Robertsb, Tatjana Nazira

aLaboratoire Langage Cerveau et Cognition, Institut des Sciences Cognitives, CNRS, 67 bd Pinel, 69675 Bron Cedex, FrancebSchool of Psychology, University of Wales, Bangor, UKcINSERM U825, Hôpital de Purpan, F-31059, Toulouse Cedex, FrancedESRC Centre for Research on Bilingualism in Theory and Practice, University of Wales, Bangor, UK

A R T I C L E I N F O

⁎ Corresponding author. L2C2, Institut des SciE-mail address: clmartin@isc.cnrs.fr (C.D.

0006-8993/$ – see front matter © 2007 Elsevidoi:10.1016/j.brainres.2007.09.049

A B S T R A C T

Article history:Accepted 12 September 2007Available online 23 October 2007

According to the ‘bilateral representation theory’, a complete copy of the words presentedfoveally is received simultaneously in the left and right visual cortices. However, a growingbody of observations, which has led to the ‘split fovea theory’, proposes a functional split ofthe foveal area between the two hemispheres. In the present study we tested these twoaccounts using an adapted version of the Reicher–Wheeler paradigm. Ten controlparticipants and ten participants with developmental dyslexia undergoingelectroencephalographic recordings were asked to identify one of five letters in a string.The target letter was systematically presented at fixation but the horizontal positioning ofthe letter string was varied such that the stimulus fluctuated in both the visual hemifieldsover the experiment. ERP results showed that letter strings encompassing the foveal fieldwere not sent to both cerebral hemispheres simultaneously when fixation coincided withextreme letter positions (i.e., first or last). Indeed, the P1 peakwas delayed in this case, whichwas interpreted as the result of a transfer of visual information from the contralateralhemisphere via the spleniumof the corpus callosum. Consistentwith the ‘split fovea theory’,this result suggests that aminimal amount of graphic input is necessary to induce a P1 event.The interhemispheric transfer time (IHTT) deducted from peak-to-peak P1 latency delaysranged from 26 to 42 ms. As previously observed, the IHTT was significantly faster for right-to-left than left-to-right transfer in the control group. IHTTwasmarginally shorter in controlparticipants as compared to participants with developmental dyslexia, and the fastertransfer to the left hemisphere seen in the former was not found in the latter.

© 2007 Elsevier B.V. All rights reserved.

Keywords:Split foveaEvent-related potentialsLetter identificationDevelopmental dyslexiaVisual fieldsVisual cortex

1. Introduction

The division of the brain in two halves creates a challenge forexplaining how foveally presented words are perceived (Brys-baert, 2004). According to a classic theoretical stand, two com-plete copies of a foveally presented visual stimulus are sent inparallel to the left and the right hemisphere (Leventhal et al.,

ences Cognitives, CNRS, 6Martin).

er B.V. All rights reserved

1988; Stone et al., 1973; Trauzettel-Klosinski and Reinhard,1998). This “bilateral representation” theory assumes that leftand right visual fields (LVF and RVF) overlap along the verti-cal meridian and that a copy of visual information presentedfoveally is sent to the primary visual cortex of each of thehemispheres. Both hemispheres then process the sameinformation without the need for interhemispheric transfer.

bd Pinel 69675 Bron Cedex, France. Fax: +33 4 37 91 12 10.

.

213B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

Support for this theory comes from the fact that patients withhemianopia sometimes show sparing of central vision (seeTrauzettel-Klosinski and Reinhard, 1998), and that horseradishperoxidase injections in the dorsal lateral geniculate nucleusshow incomplete crossing of the nasal fibers in the opticchiasm (Leventhal et al., 1988; Stone et al., 1973). However,other authors have questioned this hypothesis becausesupporting behavioral evidence is frail (e.g., Brysbaert, 1994a,b, 2004; Corballis andTrudel, 1993; Fendrich et al., 1996; Lavidorand Ellis, 2003; Lavidor et al., 2004; Celesia et al., 1993; Chiang etal., 2004; Ellis et al., 2005; Gray et al., 1997; Sugishita et al., 1994;Symonds and Mackenzie, 1957; Tootell et al., 1988). CorballisandTrudel (1993), for instance, failed to findevidence for fovealword recognition in a split-brain patient, although his perfor-mance was good for parafoveal word presentation in both LVFand RVF. Fendrich et al. (1996), who also tested split-brainpatients, suggested that each hemisphere may have a weakrepresentation of the contralateral hemi-retina, which doesnot allow fast recognition of small letters. Finally, Lavidor andWalsh (2003) showed that unilateral repetitive transcranialmagnetic stimulation (rTMS) significantly impairs lexicaldecision latencies to centrally presented words. This observa-tion clearly supports the notion that representation of foveallydisplayed words must be split between the cerebral hemi-spheres (the so-called split fovea theory) because fovealrepresentation of the non-stimulated hemisphere could oth-erwise serve recognition (see Shillcock et al., 2000).

Fig. 1 – Stimulus display and predictions from the two competinthe different experimental conditions. The target letter was alwasecond, third, fourth or fifth letter of the letter string (LP=letter pthe bilateral representation theory regarding the projection of fo

The question on the representation of foveally displayedwords in the two hemispheres of the brain leads naturally tothe study of interhemispheric transfer of information. Basedon studies of epileptic patients who have undergone callosot-omy, it is established that visual information is transferredfrom one visual cortex to its contralateral homologue via thesplenium of the corpus callosum (e.g., Censori et al., 1989). Thetime needed for this transfer of information is referred to asinterhemispheric transfer time (IHTT). IHTT in humans wasfirst estimated from reaction times in behavioral studies(Brizzolara et al., 1994; Davidson et al., 1990; Poffenberger,1912; see Bashore, 1981 for a review). In the last two decades,however, event-related potentials (ERPs) have improved theprecision of IHTT evaluation based on latencymeasures of theP1 component, a positive going ERPwavepeaking about 100msafter visual stimulus onset believed to reflect feature extrac-tion in visual areas (Tarkiainen et al., 2002). The P1 is generallyslightly delayed and smaller in amplitude over the hemisphereipsilateral to the visual field inwhich the stimulus is presented(Bayard et al., 2004). The difference in peak latencies betweenthe P1s recorded over ipsilateral and contralateral regions ofthe scalp vis-à-vis the stimulated visual field is believed toreflect callosal transfer time (see Saron and Davidson, 1989 fora review). IHTT has been estimated to be about 10–15 ms(Brown et al., 1998; Saron and Davidson, 1989), with a slowerleft-to-right than right-to-left transfer in right-handed sub-jects (Ipata et al., 1997; Saron and Davidson, 1989).

g theories. (a) Position of the stimulus vis-à-vis fixation inys centered at fixation and coincided with the first,osition). (b) Predictions from the split fovea theory andveally displayed information.

214 B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

To test the validity of the bilateral representation and splitfovea theories, we reanalyzed ERP data reported by Martinet al. (2006), focusing on the time course of the P1 component.Martin et al. (2006) tested skilled readers using the Reicher–Wheeler paradigm (Reicher, 1969; Wheeler, 1970) to describethe interplay of perceptual and lexical effects during identifi-cation of letters within letter strings. In the “Reicher–Wheelerparadigm” participants are requested to identify a target letterpresented at variable position within a letter string (either aword or a non-word) using a 2-alternative forced choice (e.g.,what was the first letter in the string: a ‘t’ or a ‘c’?). In contrastto the classic Reicher–Wheeler paradigm, however, Martinet al. (2006) presented the letter strings at five differenthorizontal positions such that the target letter was alwayspresented at fixation but could be in any of five positionswithin each string (Fig. 1a). Visual information from the entireletter string thus encompassed the foveal region of thehorizontal meridian and fluctuated in both hemifields overtrials. Since forced choice had to be performed on the targetletter and since only correct responses were included in theanalysis, this paradigm guaranteed that visual attentionwas maintained at fixation and that letter identification wasexplicit.

Depending on the positioning of the letter string vis-à-viscentral fixation, the bilateral representation theory and thesplit fovea theory make different predictions regarding thetime course of P1 over the two hemispheres. Taking intoconsideration that the fovea spans approximately 2–3° ofvisual angle, only a subpart of a letter string was perceivedfoveally in Martin et al. (2006; see Fig. 1b and Experimentalprocedure). If the fovea is representedbilaterally, the latencyofthe P1 peak over each hemisphere should be identicalindependently of stimulus positioning, given that visualinformation is always present at fixation. By contrast, thesplit fovea theory predicts that P1 peak latencies should beindistinguishable over the two hemispheres only whenfixation falls within the letter string (because visual informa-tion within each half of the fovea would elicit a P1 over thecontralateral hemisphere). Indeed, when the target letter is infirst or last position (i.e., when all but the fixated letter fall in

Fig. 2 – Mean forced choice error rates in (a) control participantsresults in the four main conditions (W50, words displayed for 50displayed for 50 ms; NW66, non-words displayed for 66 ms), as

the RVF or LVF, respectively) an “immediate” P1 should occuronly over the hemisphere contralateral to the stimulatedvisual field. Furthermore, if callosal transfer then intervenes, a“delayed” P1 should be registered over the hemisphereipsilateral to the stimulated hemifield at a later time. Here,these predictions are tested in skilled readers (control group)and in participants with developmental dyslexia (matched forage and level of education) so as to determine whether lowlevel visual processing stages indexed by P1 already distin-guish between the two groups (Walker et al., 2001).

The goal of the present study was thus to compare thepredictions of the bilateral representation theory and the splitfovea theory with respect to early visual processing duringword recognition. Lexical effects that take place in later stagesof processing have been reported in Martin et al. (2006) and arenot discussed in the present paper.

2. Results

2.1. Behavioral results

Mean forced choice error rates for the two groups of partici-pants are given in Fig. 2. A 2 between-subject by 2 by 2 by 5within-subject repeated measures ANOVA with group (con-trol/dyslexia), lexicality (word/non-word), displayduration (50/66 ms) and letter position (LP1/2/3/4/5) as factors revealed asignificant effect of lexicality (F[1,18]=14.9, p=0.001), a signif-icant effect of display duration (F[1,18]=18.2, p=0.0005), asignificant effect of letter position (F[2.82,50.75] = 9.14,pb0.0001) but no group effect (F[1,18]=3.54, p=0.08). Thegroup×lexicality interaction was significant (F[1,18]=5.25,p=0.03) and the post hoc analysis revealed a significant lexi-cality effect in control participants (p=0.004) but not in parti-cipants with developmental dyslexia (p=0.75). The letterposition×lexicality interaction was significant (F[3.05,54.93]=9.96, pb0.0001) but the group×letter position interaction wasnot (F[2.82,50.75]=1.02, p=0.39).

We observed a word superiority effect (i.e., better perfor-mance for letters embedded in words than in non-words) in

and (b) participants with developmental dyslexia. Behavioralms; W66, words displayed for 66 ms; NW50, non-words

a function of the five letter positions (LP1, 2, 3, 4 and 5).

215B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

control participants but not participants with developmentaldyslexia. Moreover, the typical viewing position effect (i.e.,better performance for words fixated near the center as com-pared to the beginning or the end) was obtained for both thegroups (significant letter position×lexicality interaction with-out group×letter position interaction). The viewing positioneffect shows that letters of the word have been integrated forlexical access (see Nazir et al., 1992, 1998, 2004 for theoreticalconsiderations about the mechanisms underlying the viewingposition effect in words).

2.2. P1 peak analysis

ERPs over the left and right parietooccipital scalp regions areshown in Fig. 3. Data are given separately for the five letterpositions within the string. There was no significant correla-tion between the latency of the P1 and reaction times oraccuracy, suggesting that this early ERP component is unre-lated to higher level lexical processes. Given that our analysesfocused on early visual processes the behavioral data were notanalyzed further.

In both groups of participants, preliminary analysesshowed that the latency of the P1 was not affected by displayduration (50 vs. 66 ms) or stimulus type (word vs. non-words).Data were therefore collapsed over these conditions andanalyzed together. The ANOVA revealed a significant effectof group (F[1,18]=8.71; p=0.01), which was due to an overall

Fig. 3 – Event-related potential results in (a) control participantsmeasured over left (PO7) and right (PO8) parietooccipital regionsstring (LP1–LP5). The arrows indicate P1 latencies in LP1 (purple)peaks in control participants over the left parietooccipital region

delay of the P1 peak latency (∼20–25 ms) in participantswith developmental dyslexia. There was no effect of electrode(F[1,18]=0.01; p=0.92) but a significant effect of letter position(F[4,15]=26.58; pb0.001), and a significant interaction betweenletter position and electrode(F[4,15]=31.66; pb0.001). This interaction reflects the fact thatthe latency of the P1 elicited by letter strings fixated at the firstor last letter (i.e., when all but the fixated letter are displayed inone visual hemifield only, see Fig. 1) was significantly delayedover the hemisphere ipsilateral to the visual field in which thestimulus was displayed. In the control group, this delay wasmost visible over left parietooccipital electrodes, where the P1peaked at 123±14 ms on average for letter position 5, vs. 98±13ms, 96±14ms, 97±14 and 98±13ms for letter positions 1, 2,3 and 4, respectively. In participants with developmentaldyslexia P1 peak latencies were 124±19 ms, 122±16 ms, 122±19 ms, 125±18 ms and 151±17 ms for letter positions 1, 2, 3, 4and 5, respectively. A mirror-reversed pattern was found overthe right parietooccipital electrodes, where the latency of theP1 was significantly delayed for letter position 1 as comparedto other letter positions, in both the participant groups.

We used Independent Component Analysis (ICA) to testwhether differences in P1 peak latencies could reflect contri-bution of different underlying components (see Experimentalprocedure). The ICA conducted on the appended series ofgrand averages obtained in all experimental conditions in thecontrol group identified a common component (#3) which

and (b) participants with developmental dyslexia. ERPsas a function of the position of the target letter in the letterand LP5 (red) conditions. Note the shorter delay between P1.

216 B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

matched the latencyof theP1measuredat PO4andgeneratedasimilar topographical representation in all conditions. Thetime course and topography of the extracted component 3were compared to the time course of grand-average ERPs andP1 topographies in each of the experimental conditions (Fig. 4).Subtraction of independent component 3 from individual ERPsin 8 of the control participants (Klein and Feige, 2005) followedby a direct comparison of peak mean amplitudes in the timerange of the P1 (60–160 ms) revealed no significant differencesbetween fixation positions 1 and 5 in any of the experimentalconditions (all psN0.1).

2.3. Interhemispheric transfer time (IHTT)

Based on Saron andDavidson (1989), we estimated left-to-rightIHTT by subtracting P1 peak latencies measured over the rightand left parietooccipital regions (PO7 and PO8) for letterposition 1. Right-to-left IHTT was estimated by subtracting P1peak latencies over the left and the right parietooccipitalregions for letter position 5. Note that lateral occipital sites arethought to provide a better evaluation of IHTT than themedialsites (Saron andDavidson, 1989). Themeanestimated right-to-left IHTT was 26±14 ms in control participants and 42±17 msin participantswith developmental dyslexia. Themean left-to-right IHTTwas 42±15ms in control participants and 40±10msin participants with developmental dyslexia. The difference inestimated IHTT between the two groups was marginallysignificant (F[2,17]=4.51; p=0.06). Measure of effect size (etasquare value) revealed that 36% of the variance in the depen-dent variable could be attributed to the group factor. The IHTTof control participants was significantly faster for right-to-leftthan left-to-right transfer (t[9]=2.654; p=0.026). In participantswith developmental dyslexia IHTT did not differ in the twoconditions (t[9]=− .284; p=0.783).

Fig. 4 – Independent Component Analysis of ERPs in conditionsconditions for LP1 (purple) and LP5 (red) with activity produced bof topographic maps of the P1 and ICA component 3 in the vario

3. Discussion

The present results characterized an early positive ERP event(P1) over occipitoparietal regions contralateral to the stimulat-ed visual hemifield. Peak latency was approximately 100 msafter stimulus onset in control participants and 123 ms inparticipants with developmental dyslexia. This P1, which isbelieved to reflect processes in early visual areas that areinvolved in feature extraction (Tarkiainen et al., 2002) –irrespective of the linguistic category of the stimulus (Rossionet al., 2003) –wasunaffected bydisplayduration (50 or 66ms) orstimulus type (word or non-word).

3.1. P1 peak latency and the split fovea hypothesis

When participants fixated a letter situated within the letterstring (i.e., letter positions 2, 3, and 4) the P1 peaked at aboutthe same latency over the left and right parietooccipital scalp.However, when participants fixated the first or the last letter inthe letter string (letter positions 1 and 5), i.e., when all but oneletter fell into one visual hemifield, P1 peak latency wasdelayed by 26 to 42 ms over the ipsilateral scalp. Critically, theextraction of independent components in the P1 range in thedifferent conditions using ICA lead to the characterization of asingle component accounting for the P1 for both fixation po-sitions 1 and 5 (Fig. 4). Therefore, the difference in P1 peaklatency genuinely reflects a shift in latency of the P1 generatorsand cannot be attributed to differential contribution of over-lapping components in the P1 range.

This delay, which was observed in both the groups, isinconsistent with the view that identical copies of stimuliencompassing the foveal field are sent simultaneously to eachof the two cerebral hemispheres. Our result is thus consistent

LP1 and LP5. (a) Grand-average ERPs in all experimentaly ICA component 3 superimposed (dotted). (b) Comparisonus experimental conditions for letter positions 1 and 5.

217B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

with the split fovea theory (Brysbaert, 1994b, 2004; Ellis et al.,2005; Lavidor et al., 2004) and go against the predictions of thebilateral representation theory (Leventhal et al., 1988; Stoneet al., 1973; Trauzettel-Klosinski and Reinhard, 1998).

The present data also show that a minimal amount ofgraphic input is necessary (∼1.5 letters in this case, i.e., theamount of visual input to one hemifield when the letter stringis fixated at position 2 or 4; see Fig. 1) to elicit a P1 over thecontralateral occipitoparietal scalp. When less than 1.5 lettersis presented in each hemifield (i.e., when fixation coincideswith the first or last letter in the string), no immediate con-tralateral P1 is observed. In addition, beyond1.5 letters, P1 peaklatency remains constant independently of the number ofadditional letters displayed in the contralateral hemifield.Given that, in this experiment, 1.5 letters spanned approxi-mately 2° of visual angle, the contralateral P1 found forpositions 2 and 4 is probably triggered by visual informationperceived on and in the proximity of the fovea.

It is important to stress here again that the P1 componentprobably reflects processes of visual feature extraction (Tar-kiainen et al., 2002), which are not distinguishable for wordsand non-words. Therefore, the present results only enable usto infer that information about foveally displayed visualstimuli is not processed in parallel and redundantly in eachof the two hemispheres. Speculations on how visual informa-tion from the two visual fields is subsequently integrated inthe process of visual word recognition cannot be inferreddirectly from our results and await dedicated investigations.In this context it is worth noting that Hunter et al. (2007)recently investigated visual word recognition in participantswith typical (left) and atypical (right) language lateralizationusing the same variable viewing position paradigm as thatused in the present study. Their results showed that languagedominance has a significant impact on the way letters in thetwo visual fields are processed, with an advantage for lettersthat are directly sent to the language dominant hemisphere.This result was found even for foveally fixated words, whichsupports our interpretation that foveally displayed stimuli aresplit between the two hemispheres and benefit differentlyfrom higher level processes.

3.2. Interhemispheric transfer

We found that visual information is transferred from thecontralateral to the ipsilateral hemisphere with a delay of 26 to42 ms in control participants and 40 to 42 ms in participantswith developmental dyslexia. Note that previous studies haveestimated IHTT to about 10–15 ms (Barnett and Corballis, 2005;Brown et al., 1998, 1994; Murray et al., 2001; Saron andDavidson, 1989). The longer delays observed in our experimentmay be explained by the fact that our stimuli were letter strings,which are substantiallymore complex than basic visual stimulitypically used to measure IHTT (e.g., checkerboards or flashes).In addition, the delays measured here are consistent with theargument put forward by Ono et al. (2002) that interhemispher-ic transfer should theoretically exceed axonal conduction time,which is estimated to take about 20 ms.

Several studies in healthy participants have shown thatright-to-left interhemispheric transfer is typically faster thanleft-to-right transfer (Brown and Jeeves, 1993; Brown et al.,

1994; Ipata et al., 1997; Marzi et al., 1991; Saron and Davidson,1989). Threemain explanationshave beenproposed to accountfor this asymmetry: (i) according to Marzi et al. (1991) thedifference could relate to callosal connection asymmetries,with a greater number of neurons projecting from the right tothe left hemisphere than vice versa (see also Saron andDavidson, 1989). (ii) A second possibility is that IHTT issystematically shorter when information is transferred fromthe non-dominant to the dominant hemisphere — in thepresent case, the left hemisphere (Nowicka et al., 1996).However, a meta-analysis of ERP studies by Brown et al.(1994) did not support this claim. (iii) Third,Miller (1996) arguedthat the right hemisphere has more myelinated axonsenabling rapid conduction and processing of gestalt represen-tations. In line with Miller (1996), Barnett and Corballis (2005)and Barnett and Kirk (2005) suggested that the presence offaster conducting cortico-cortical myelinated axons in theright hemisphere increases the likelihood of neural summa-tion in that hemisphere, and therefore the speed of right-to-left relative to left-to-right transfer. While IHTT measures incontrol participants replicated this asymmetry, our results donot enable us to validate or invalidate any of the threehypothetical explanations above.

3.3. Comparison of control participants and participantswith developmental dyslexia

Even though the ANOVA on P1 latencies revealed a significantgroup effect, the group×letter position×electrode interactionwas not significant. Therefore, interpretations arising fromcomparisons between groupsmust be cautious. The P1 peakedsignificantly earlier in control participants than in participantswith developmental dyslexia, indicating that differences canbe found in early visual processing stages. IHTT was alsomarginally faster in control participants than in participantswith developmental dyslexia and this differencewas driven bythe shorter right-to-left IHTT in control participants. Since ourfinding is only a marginally significant effect, it is not whollyinconsistent with results reported by other investigators (e.g.,Davidson et al., 1990), who showed no overall IHTT groupdifferences. In our study, the IHTT difference between controlparticipants and participants with developmental dyslexiawas smaller than 11 ms. Such a small difference might havebeen undetectable with the sample size of Davidson et al.(1990). The absence of asymmetry between left-to-right andright-to-left IHTT in participants with developmental dyslexiamay indicate a reduction of white matter density in the lefthemisphere, as hypothesized by several authors (Beaulieuet al., 2005; Deutsch et al., 2005; Silani et al., 2005). It is worthnoting that global abnormalities of the corpus callosum –whichhave beendescribed inparticipantswith developmentaldyslexia (Duara et al., 1991) – cannot account for the presentfindings because the difference between control participantsand participants with developmental dyslexia was only ob-served for transfers to the language dominant hemisphere,and not the non-dominant hemisphere. In line with Barnettand Corballis (2005) and Barnett and Kirk (2005), we hypoth-esize that abnormalities of the myelinated pathways in theright hemisphere of participants with developmental dyslexiamight account for the slower right-to-left transfer time. Recall

Fig. 5 – Experimental design of the forced two-choice task

218 B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

though that neuroimaging studies have identified abnormal-ities in the left rather than the right hemisphere in develop-mental dyslexia (Beaulieu et al., 2005; Deutsch et al., 2005;Silani et al., 2005). Even though hemispheric dominance doesnot necessarily play a role in the asymmetry of IHTT in controlparticipants, the atypical hemispheric dominance for languageprocessing often reported in developmental dyslexia mayrelate to the lack of IHTT asymmetry reported here (see forinstance Eckert et al., 2003, 2005; Giraud et al., 2005; McCroryet al., 2000; Simos et al., 2000, 2002). Together with P1 peaklatency results, the present findings suggest that early visualprocessing stages are affected in developmental dyslexia andmay thus contribute to associated reading difficulties.

(adapted version of the Reicher–Wheeler paradigm).

4. Experimental procedure

The experimental procedure of this paper is largely based onMartin et al. (2006) where additional details can be found.

4.1. Participants

Ten control participants (5 females and 5 males; mean age24.8±1.6 years) and ten participants with developmentaldyslexia (2 females and 8 males; mean age 25.2±6.3) gaveinformed consent to participate in the experiment that wasapproved by a local ethics committee. They were all right-handed native speakers of French, with normal global IQ(assessed using the WAIS-R), and normal or corrected-to-nor-mal vision. We recruited volunteers who had been diagnosedas dyslexic and had documented histories of reading andspelling difficulties. Participants were tested with tasks de-signed to assess reading accuracy and speed, and phonologicalprocessing: reading of words and non-words, rapid digit nam-ing and phoneme awareness tasks (phoneme deletion, rhymejudgment and spoonerisms; see Table 1).

4.2. Stimuli

Stimuli were 40 French nouns selected from the Brulexdatabase (Content and Radeau, 1990) and 40 unpronounceable

Table 1 – Participant performance in intelligence, readingand phonological tasks in the two groups of participants

Controlparticipants

Dyslexicparticipants

P

Full IQ 118 (11) 109 (8) n.s.Verbal IQ 119 (12) 109 (13) n.s.Performance IQ 111 (13) 107 (14) n.s.Word reading 533 (74) 740 (78) ⁎Non-word reading 660 (140) 919 (114) ⁎Rapid digit naming 15 (2) 22 (4) ⁎Phoneme deletion 99 (19) 241 (49) ⁎⁎Rhyme judgment 1220 (256) 1826 (166) ⁎⁎Spoonerisms 96 (32) 229 (79) ⁎⁎

Standard deviations are shown in parentheses. Average scores andstandarddeviations refer tomarks fromnormalizedneuropsychologicaltests. n.s.=Not significant; ⁎Pb0.05; ⁎⁎Pb0.01.

consonant strings (e.g., ‘PGSRF’). Both words and non-wordswere 5 letters long.

4.3. Task and procedure

At the beginning of each trial, a fixation cross was displayed atthe center of a computer screen for 2.5 s. The fixation thendisappeared for 50 ms before a second fixation appeared for200ms. After a secondpause of 50ms, a stimulus (word or non-word) was displayed for either 50 or 66 ms in “Times NewRoman” lower case, font size 28. The stimulus position wasvaried laterally so that each of the five letters in a string couldcoincide with fixation. The amount of visual informationtherefore fluctuated across foveal and parafoveal fieldsbetween the experimental conditions (Fig. 1). Stimuli sub-tended 6.65° of visual angle, at a distance of 60 cm. After 50 or66 ms, the stimulus was replaced by a string made of 7 uppercase Xs masking the entire stimulus string (mask) and twoprobe letters, one above and one below themask. Subjects hadto indicate which of the two probe letters was the letterpreviously presented at fixation (target letter), by pressing thetop or bottom button of a response pad (Fig. 4). The mask andthe two probe letters remained on the screen until theresponse. To increase statistical power, each of the 40 wordsand non-words were presented 6 times in the 50-ms conditionand 6 times in the 66-ms condition. Overall, the 960 trials werepseudo-randomly distributed in 12 blocks of 80, with each itemdisplayed only once per block. Exposure durations (50 and66 ms) and stimulus categories (words and non-words) wererandomized within each block. Block order and response sidewere counterbalanced across participants (Fig. 5).

4.4. ERP acquisition and processing

Participants were comfortably seated in a quiet room andasked to refrain from moving and blinking. Electrophysiolog-ical data were recorded from 64 Ag/AgCl electrodes (placedaccording to the extended International 10–20 system) at asampling rate of 500 Hz, using SynAmps™ amplifiers (Neu-roscan™, El Paso, TX, USA). The electrooculogram wasrecorded using supraorbital and infraorbital electrodesconnected to a bipolar channel. Signals were filtered on-linebetween 0.1 and 100 Hz. Impedances were kept below 20 kΩ.Continuous recordings were digitally band-pass filtered off-

219B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

line in the interval [1–40] Hz. Eye-blink artefacts weremathematically corrected and remaining artefacts manuallydismissed. Epochs ranged from −100 to 1000ms after the onsetof the five letter string (word or non-word). Signal variationsexceeding ±75 μm anywhere in the analysis window and onany of the channels except the vertical electrooculogram(VEOG) were automatically dismissed. After baseline correc-tion relative to pre-stimulus activity and rejection of errors,there were at least 48 epochs per condition in all participants.Individual difference waveforms and grand-average wave-forms were then derived from individual ERPs.

4.5. Statistical analysis

P1 peak search was confined to the interval 70–130 ms basedon the corresponding period of maximal activity in the MeanGlobal Field Power over the 64 electrodes (Picton et al., 2000).P1 latencies at the electrodes of maximal sensitivity wereanalyzed using an analysis of variance (ANOVA) with group(control vs. participantswith developmental dyslexia) as inter-subject variable and electrode (over left and right occipitotem-poral regions) and letter position (LP1 to LP5) as intra-subjectvariables. An “extended” Independent Component Analysis(ICA), using the infomax algorithm (Makeig et al., 1997), wasjointly applied to the appended averages of all experimentalconditions in order to separate generators associated withdifferent topographies. Visual inspection of both temporal andtopographic representations of the ICA matrices allowedidentification of one component best accounting for the P1(see Results). The activity produced by the component wasthen subtracted from individual ERPs in all conditions in 8 ofthe 10 control participants and ERP mean amplitudes werecompared in the P1 range between conditions (Klein and Feige,2005). Two participants who failed to display a clear compo-nent following ICA were excluded from this analysis.

Acknowledgments

The authors wish to thank Yves Paulignan, Chantal BlanchardandNicolas Chauveau for the useful discussions and technicalassistance.

R E F E R E N C E S

Barnett, K.J., Corballis, M.C., 2005. Speeded right-to-leftinformation transfer: the result of speeded transmission inright-hemisphere axons? Neurosci. Lett. 380, 88–92.

Barnett, K.J., Kirk, I.J., 2005. Lack of asymmetrical transfer forlinguistic stimuli in schizophrenia: an ERP study. Clin.Neurophysiol. 116, 1019–1027.

Bashore, T.R., 1981. Vocal and manual reaction time estimates ofinterhemispheric transmission time. Psychol. Bull. 89, 352–368.

Bayard, S., Gosselin, N., Robert, M., Lassonde, M., 2004. Inter- andintra-hemispheric processing of visual event-relatedpotentials in the absence of the corpus callosum. J. Cogn.Neurosci. 16, 401–414.

Beaulieu, C., Plewes, C., Paulson, L.A., Roy, D., Snook, L., Concha, L.,et al., 2005. Imaging brain connectivity in children with diversereading ability. NeuroImage 25, 1266–1271.

Brizzolara, D., Ferretti, G., Brovedani, P., Casalini, C., Sbrana, B.,1994. Is interhemispheric transfer time related to age? Adevelopmental study. Behav. Brain Res. 64, 179–184.

Brown, W.S., Jeeves, M.A., 1993. Bilateral visual field processingand evoked potential interhemispheric transmission time.Neuropsychologia 31, 1267–1281.

Brown, W.S., Larson, E.B., Jeeves, M.A., 1994. Directionalasymmetries in interhemispheric transmission time: evidencefrom visual evoked potentials. Neuropsychologia 32, 439–448.

Brown, W.S., Bjerke, M.D., Galbraith, G.C., 1998. Interhemispherictransfer in normals and acallosals: latency adjusted evokedpotential averaging. Cortex 34, 677–692.

Brysbaert, M., 1994a. Behavioral estimates of interhemispherictransmission time and the signal detection method: areappraisal. Percept. Psychophys. 56, 479–490.

Brysbaert, M., 1994b. Interhemispheric transfer and theprocessing of foveally presented stimuli. Behav. Brain Res. 64,151–161.

Brysbaert, M., 2004. The importance of interhemispheric transferfor foveal vision: a factor that has been overlooked in theoriesof visual word recognition and object perception. Brain Lang.88, 259–267.

Celesia, G.G., Meredith, J.T., Pluff, K., 1983. Perimetry, visualevoked potentials and visual evoked spectrum array inhomonymous hemianopsia. Electroencephalogr. Clin.Neurophysiol. 56, 16–30.

Censori, B., Del Pesce, M., Provinciali, L., Quattrini, A., Mancini, S.,Papo, I., 1989. Functions of the corpus callosum: observationsfrom callosotomy performed for intractable epilepsy. Boll. Soc.Ital. Biol. Sper. 65, 53–59.

Chiang, T.C., Walsh, V., Lavidor, M., 2004. The corticalrepresentation of foveal stimuli: evidence fromquadrantanopia and TMS-induced suppression. Brain Res.Cogn. Brain Res. 21, 309–316.

Content, A.M., Radeau, P., 1990. M. BRULEX: une base de donnéeslexicales informatisée pour le français écrit et parlé. AnnéePsychol. 90, 551.

Corballis, M.C., Trudel, C.I., 1993. Role of the forebraincommissures in interhemispheric integration.Neuropsychology 7, 306–324.

Davidson, R.J., Leslie, S.C., Saron, C., 1990. Reaction time measuresof interhemispheric transfer time in reading disabled andnormal children. Neuropsychologia 28, 471–485.

Deutsch, G.K., Dougherty, R.F., Bammer, R., Siok,W.T., Gabrieli, J.D.,Wandell, B., 2005. Children's reading performance is correlatedwith white matter structure measured by diffusion tensorimaging. Cortex 41, 354–363.

Duara, R., Kushch, A., Gross-Glenn, K., Barker, W.W., Jallad, B.,Pascal, S., et al., 1991. Neuroanatomic differences betweendyslexic and normal readers on magnetic resonance imagingscans. Arch. Neurol. 48, 410–416.

Eckert, M.A., Leonard, C.M., Richards, T.L., Aylward, E.H.,Thomson, J., Berninger, V.W., 2003. Anatomical correlates ofdyslexia: frontal and cerebellar findings. Brain 126,482–494.

Eckert, M.A., Leonard, C.M., Wilke, M., Eckert, M., Richards, T.,Richards, A., et al., 2005. Anatomical signatures of dyslexia inchildren: unique information from manual and voxel basedmorphometry brain measures. Cortex 41, 304–315.

Ellis, A.W., Brooks, J., Lavidor, M., 2005. Evaluating a split foveamodel of visual word recognition: effects of case alternation inthe two visual fields and in the left and right halves of wordspresented at the fovea. Neuropsychologia 43, 1128–1137.

Fendrich, R.,Wessinger, C.M., Gazzaniga, M.S., 1996. Nasotemporaloverlap at the retinal vertical meridian: investigations with acallosotomy patient. Neuropsychologia 34, 637–646.

Giraud, K., Demonet, J.F., Habib, M., Marquis, P., Chauvel, P.,Liegeois-Chauvel, C., 2005. Auditory evoked potential patternsto voiced and voiceless speech sounds in adult developmental

220 B R A I N R E S E A R C H 1 1 8 5 ( 2 0 0 7 ) 2 1 2 – 2 2 0

dyslexics with persistent deficits. Cereb. Cortex 15,1524–1534.

Gray, L.G., Galetta, S.L., Siegal, T., Schatz, N.J., 1997. The centralvisual field in homonymous hemianopia. Evidence forunilateral foveal representation. Arch. Neurol. 54, 312–317.

Hunter, Z.R., Brysbaert, M., Knecht, S., 2007. Foveal word readingrequires interhemispheric communication. J. Cogn. Neurosci.19 (8), 1373–1387.

Ipata, A., Girelli, M., Miniussi, C., Marzi, C.A., 1997.Interhemispheric transfer of visual information in humans:the role of different callosal channels. Arch. Ital. Biol. 135,169–182.

Klein, C., Feige, B., 2005. An Independent Component Analysis(ICA) approach to the study of developmental differences in thesaccadic contingent negative variation. Biol. Psychol. 70,105–114.

Lavidor, M., Ellis, A.W., 2003. Interhemispheric integration ofletter stimuli presented foveally or extra-foveally. Cortex 39,69–83.

Lavidor, M., Walsh, V., 2003. A magnetic stimulation examinationof orthographic neighborhood effects in visual wordrecognition. J. Cogn. Neurosci. 15, 354–363.

Lavidor, M., Hayes, A., Shillcock, R., Ellis, A.W., 2004. Evaluating asplit processing model of visual word recognition: effects oforthographic neighborhood size. Brain Lang. 88, 312–320.

Leventhal, A.G., Ault, S.J., Vitek, D.J., 1988. The nasotemporaldivision in primate retina: the neural bases of macular sparingand splitting. Science 240, 66–67.

Makeig, S., Bell, A.J., Jung, T.P., Ghahremani, D., Sejnowski, T.J.,1997. Blind separation of auditory event-related brainresponses into independent components. Proc. Natl. Acad. Sci.U. S. A. 94, 10979–10984.

Martin, C.D., Nazir, T., Thierry, G., Paulignan, Y., Demonet, J.F.,2006. Perceptual and lexical effects in letter identification: anevent-related potential study of the word superiority effect.Brain Res. 1098, 153–160.

Marzi, C.A., Bisiacchi, P., Nicoletti, R., 1991. Is interhemispherictransfer of visuomotor information asymmetric? Evidencefrom a meta-analysis. Neuropsychologia 29, 1163–1177.

McCrory, E., Frith, U., Brunswick, N., Price, C., 2000. Abnormalfunctional activation during a simple word repetition task: aPET study of adult dyslexics. J. Cogn. Neurosci. 12, 753–762.

Miller, R., 1996. Axonal Conduction Time and Human CerebralLaterality: A Psychobiological Theory. Harwood Academic,Amsterdam.

Murray, M.M., Foxe, J.J., Higgins, B.A., Javitt, D.C., Schroeder, C.E.,2001. Visuo-spatial neural response interactions in earlycortical processing during a simple reaction time task: ahigh-density electrical mapping study. Neuropsychologia 39,828–844.

Nazir, T.A., Heller, D., Sussmann, C., 1992. Letter visibility andword recognition: the optimal viewing position in printedwords. Percept. Psychophys. 52, 315–328.

Nazir, T.A., Jacobs, A.M., O'Regan, J.K., 1998. Letter legibility andvisual word recognition. Mem. Cogn. 26, 810–821.

Nazir, T.A., Ben-Boutayab, N., Decoppet, N., Deutsch, A., Frost, R.,2004. Reading habits, perceptual learning, and recognition ofprinted words. Brain Lang. 88, 294–311.

Nowicka, A., Grabowska, A., Fersten, E., 1996. Interhemispheric

transmission of information and functional asymmetry of thehuman brain. Neuropsychologia 34, 147–151.

Ono, T., Matsuo, A., Baba, H., Ono, K., 2002. Is a cortical spikedischarge “transferred” to the contralateral cortex via thecorpus callosum?: an intraoperative observation ofelectrocorticogram and callosal compound action potentials.Epilepsia 43, 1536–1542.

Picton, T.W., Bentin, S., Berg, P., Donchin, E., Hillyard, S.A., JohnsonJr., R., et al., 2000. Guidelines for using human event-relatedpotentials to study cognition: recording standards andpublication criteria. Psychophysiology 37, 127–152.

Poffenberger, A.T., 1912. Reaction time to retinal stimulation withspecial reference to the time lost in conduction throughnervous centers. Arch. Psychol. 23, 1–73.

Reicher, G.M., 1969. Perceptual recognition as a function ofmeaninfulness of stimulus material. J. Exp. Psychol. 81,275–280.

Rossion, B., Joyce, C.A., Cottrell, G.W., Tarr, M.J., 2003. Earlylateralization and orientation tuning for face, word, andobject processing in the visual cortex. NeuroImage 20,1609–1624.

Saron, C.D., Davidson, R.J., 1989. Visual evoked potential measuresof interhemispheric transfer time in humans. Behav. Neurosci.103, 1115–1138.

Shillcock, R., Ellison, T.M., Monaghan, P., 2000. Eye-fixationbehavior, lexical storage, and visual word recognition in a splitprocessing model. Psychol. Rev. 107, 824–851.

Silani, G., Frith, U., Demonet, J.F., Fazio, F., Perani, D., Price, C., et al.,2005. Brain abnormalities underlying altered activation indyslexia: a voxel based morphometry study. Brain 128,2453–2461.

Simos, P.G., Breier, J.I., Fletcher, J.M., Bergman, E., Papanicolaou, A.C., 2000. Cerebral mechanisms involved in word reading indyslexic children: a magnetic source imaging approach. Cereb.Cortex 10, 809–816.

Simos, P.G., Fletcher, J.M., Foorman, B.R., Francis, D.J., Castillo, E.M.,Davis, R.N., et al., 2002. Brain activation profiles during the earlystages of reading acquisition. J. Child Neurol. 17, 159–163.

Stone, J., Leicester, J., Sherman, S.M., 1973. The naso-temporaldivision of the monkey's retina. J. Comp. Neurol. 150, 333–348.

Sugishita, M., Hamilton, C.R., Sakuma, I., Hemmi, I., 1994.Hemispheric representation of the central retina ofcommissurotomized subjects. Neuropsychologia 32, 399–415.

Symonds, C., Mackenzie, I., 1957. Bilateral loss of vision fromcerebral infarction. Brain 80, 415–455.

Tarkiainen, A., Cornelissen, P.L., Salmelin, R., 2002. Dynamics ofvisual feature analysis and object-level processing in faceversus letter-string perception. Brain 125, 1125–1136.

Tootell, R.B., Switkes, E., Silverman, M.S., Hamilton, S.L., 1988.Functional anatomy of macaque striate cortex: II. Retinotopicorganization. J. Neurosci. 8, 1531–1568.

Trauzettel-Klosinski, S., Reinhard, J., 1998. The vertical field borderin hemianopia and its significance for fixation and reading.Invest. Ophthalmol. Vis. Sci. 39, 2177–2186.

Walker, M.M., Spires, H., Rastatter, M.P., 2001. Hemisphericprocessing characteristics for lexical decisions in adults withreading disorders. Percept. Mot. Skills 92, 273–287.

Wheeler, D.D., 1970. Processes in word recognition. Cogn. Psychol.1, 59–85.