Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Fast, visual specialization for reading in English revealed by the topography of the N170 ERP response

Fast, visual specialization for reading in English revealed by the topography of the N170 ERP... Background: N170 effects associated with visual words may be related to perceptual expertise effects that have been demonstrated for faces and other extensively studied classes of visual stimuli. Although face and other object expertise effects are typically bilateral or right-lateralized, the spatial topography of reading-related N170 effects are often left-lateralized, providing potential insights into the unique aspects of reading-related perceptual expertise. Methods: Extending previous research in German [1], we use a high-density channel array to characterize the N170 topography for reading-related perceptual expertise in English, a language with inconsistent spelling-to-sound mapping. N170 effects related to overall reading-related expertise are defined by contrasting responses to visual words versus novel symbol strings. By contrasting each of these conditions to pseudowords, we examined how this reading-related N170 effect generalizes to well-ordered novel letter strings. Results: A sample-by-sample permutation test computed on word versus symbol ERP topographies revealed differences during two time windows corresponding to the N170 and P300 components. Topographic centroid analysis of the word and symbol N170 demonstrated significant differences in both left-right as well as inferior-superior dimensions. Words elicited larger N170 negativities than symbols at inferior occipito-temporal channels, with the maximal effect over left inferior regions often unsampled in conventional electrode montages. Further contrasts produced inferior-superior topographic effects for the pseudoword-symbol comparison and left-lateralized topographic effects for the word-pseudoword comparison. Conclusion: Fast specialized perception related to reading experience produces an N170 modulation detectable across different EEG systems and different languages. Characterization of such effects may be improved by sampling with greater spatial frequency recordings that sample inferior regions. Unlike in German, reading-related expertise effects in English produced only partial generalization in N170 responses to novel pseudowords. The topographic inferior-superior N170 differences may reflect general perceptual expertise for orthographic strings, as it was found for words and pseudowords across both languages. The topographic left-right N170 difference between words and pseudowords was only found in English, and may suggest that ambiguity in pronunciating novel pseudowords due to inconsistency in spelling-to-sound mapping influences early stages of letter string processing. Page 1 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 One recent study of native German speaking adults dem- Background The N170 is a component of the event-related potential onstrated sublexical N170 effects by contrasting responses (ERP) peaking between 150 and 200 ms and showing an to strings of novel symbols to letters strings grouped into occipito-temporally negative and fronto-centrally positive common orthographic patterns (familiar or novel word topography. It is strongly elicited by certain classes of vis- forms in German), demonstrating that specialization of ual stimuli, such as faces [2,3], relative to other visual con- processing as assessed by either lexical or sublexical trol stimuli. Investigations of the psychological principles strings is strongest at inferior occipito-temporal channels that drive the N170 to respond more strongly to some [1]. However, the region that demonstrated the peak effect classes of stimuli over others have demonstrated perceptual was at the edge of the montage of electrodes applied, sug- expertise effects across several classes of stimuli, including gesting that this effect might be better characterized by an enhanced N170 responses (relative to other object control electrode array, which covers regions inferior to the classi- stimuli) for bird experts viewing birds [4], car experts cal 10–20 or 10-10 electrode montages [23,24]. viewing cars [5], and has even been demonstrated for lab- oratory-induced expertise with 3D novel figures ("gree- The aim of the current study was to further characterize bles" [6]). These results support a potential relationship the nature of reading-related N170 expertise effects by between extensive visual experience with a stimulus applying a 129-channel array (geodesic sensor net, Electri- domain and alterations in visual processes within the first cal Geodesics, Inc.) that extends the coverage of the classi- 200 ms of perceptual identification. This framework of cal 10-10 montage [23,24] to more inferior regions, perceptual expertise may also account for experience- thereby providing adequate spatial sampling of the peak dependent changes in reading skill – a domain in which effect of interest (Fig. 1). This study adopts a paradigm extensive practice develops considerable visual expertise used in a previous N170 study conducted with German at the level of letter-strings and the pattern by which letters speaking subjects [1] to explore potential replicability of typically are combined to create visual word forms [7]. effects across languages that differ in the level of consist- ency of how letters map onto word sounds [25], as well as Neurophysiological studies have shown that skilled adult to examine the topography of the N170 with a greater spa- readers develop fast, perceptual identification processes tial sampling of inferior regions that might be critical to that are specialized for words and other letter strings, capturing effects produced near ventral posterior brain reflected by differences in N170 responses compared to regions (see [7] for review). EEG was recorded continu- control stimuli, such as symbol strings, that control for ously as participants actively monitored for an occasional visual features [1,8-10]. target, defined as an immediately repeated item (i.e. "one- back"), among a series of word, pseudoword, or symbol Unlike findings of right-lateralized or bilateral N170 string stimuli. Advantages of this paradigm include rela- responses for faces, N170 responses to word stimuli tively equal engagement for all classes of stimuli, which showed a left-lateralized topography [1,3,11-13]. Across can be assessed via behavioral responses to targets, as well studies, however, the degree of the left-lateralization var- the segregation of response-free trials from infrequent tar- ied between strong [3,12] and moderate [1,13]. get trials within ERP analyses. Some studies also showed that the N170 is sensitive for Results Behavior linguistic processing [14,15]. Consonant strings had larger N170 amplitudes than words [14,15], and sublexi- Participants detected targets with a high accuracy across cally irregular pseudowords were in between [14]. Other all conditions (>90% in each condition, see table 1), indi- studies, however, did not find N170 differences between cating that all conditions were relatively easy. Subtle con- words and pseudowords [1,8,16]. In one study, the differ- dition differences were detectable, however, via repeated ences between consonant strings and words were only measure ANOVA analyses, which revealed a main effect of found for lexical and semantic tasks, but not during stimulus condition for accuracy (F(2,13) = 6.23, p < 0.05), implicit reading [8], whereas it was found across seman- but not for reaction time (F(2,13) = 0.61, p = ns). Symbol tic, passive, and implicit viewing in another [14]. These strings were detected slightly less accurately than the other results suggest that N170 responses are somewhat variable two conditions. across experiments, which might be due to different task demands and presentation modes. Word-symbol differences in consecutive ERP maps To assess differential processing of words and symbol Word frequency effects in the N170 were more consist- strings over time, a Topographic Analysis of Variance ently found across studies, with low frequency words pro- (TANOVA, [26]) on non-normalized (raw) ERP maps was ducing more negative N170 amplitudes [17-21] (but see computed for each time point. TANOVA on raw maps also [22]). detects all systematic amplitude differences between two Page 2 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 Table 1: Behavioral results for detecting targets Words Pseudowords Symbols Accuracy (% correct) 98.4 95.5 90.1 Fp1 Fp2 Reaction time (ms) 609 599 542 F7 F8 F3 F4 Fz for the N170 component, we averaged samples across the time segment between the two GFP minima (based on the C3 C4 Cz T3/T7 T4/T8 average of word and symbol grandmeans) that marked the beginning and end of the N170 component as in [1]. P3 P4 Pz The N170 segment maps (144–248 ms) of the word, pseudoword, and symbol conditions were subsequently 57/TP9 57/TP10 T5/P7 T6/P8 analyzed to characterize GFP and topography effects across these stimulus groups. Topographic effects were O1 O2 tested using difference map t-statistics, centroid analyses, 65 91 and analyses of selected channels. First, overall reading- 69 70/PO9 90/PO10 95 related N170 effects were analyzed, comparing words vs. symbol strings. Second, generalization of reading-related N170 specialization to novel word forms was tested com- paring pseudowords both to symbols and to words. H Figure 1 igh-density 129-channel montage High-density 129-channel montage. Filled black dots Statistical difference map analyses indicate electrodes corresponding to the 10–20 system posi- The word N170 topography showed the largest negativity tions [34]. Inferior occipito-temporal channel groups used at occipito-temporal electrodes with a maximum over the for waveform illustration and additional lateralization analy- ses are marked with dotted circles. Note that the high-den- left hemisphere and the largest positivity at fronto-central sity montage extends both the 10-10 and 10-5 montages [24] electrodes. The symbol N170 topography showed the for an additional inferior row (approx. 5% of the Nasion- largest negativity at occipito-parietal electrodes with a Inion distance). maximum over the right hemisphere and the largest posi- tivity at fronto-polar electrodes. The two topographies clearly differed, as shown by the large t-values in the sta- tistical difference map (Fig. 2). The maximal effects were found at parietal electrodes and at left inferior temporal and occipito-temporal electrodes at the edge of the chan- nel array (Fig. 2). maps (i.e. including all 129 electrodes). Accordingly, word and symbol processing differed (p < 0.01, to adjust N170 responses to pseudowords and symbols clearly dif- for multiple comparisons) during two separate time win- fered in the statistical difference map (Fig. 3). The maxi- dows, from 160–244 ms and from 324–512 ms, largely mal effects were found at parietal electrodes and overlapping with the Global Field Power (GFP [27]) peaks bilaterally at inferior electrodes at the edge of the elec- of the N170 and P300 ERP components (Fig. 2). trode array. N170 time window The statistical difference map also indicated clear N170 Our approach to analyzing topographic effects in the effects between responses to words and pseudowords (Fig. N170 time window follows an ERP mapping approach 3). These effects were left-lateralized, as they were found [28,29], designed to take full advantage of information at many occipito-temporal channels over the left hemi- from all the channels in the high-density channel array. sphere, but hardly over the right hemisphere. The pseu- According to this approach, ERPs are seen as a series of doword N170 topography resembled the word N170, but maps changing in Global Field Power (GFP [27]) and showed a more bilateral negativity and a positivity cen- topography over time. Moreover, ERP topographies tend tered around the fronto-central midline electrodes, in to remain stable for short periods of time, typically chang- contrast to the more left-lateralized negativity and positiv- ing at time points with low GFP. To get a robust measure ity of the word N170 (Fig. 3). Page 3 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 TANOVA (word-symbol difference) GFP words p<0.01 symbols N170 P300 p<0.05 0 -100 0 100 200 300 400 500 600 700 800ms symbol N170 t-map word N170 Gener wo Figure 3 rd fa olization of reading rms -related N170 expertise to novel -2 +2 0µV +6 -6 0t Generalization of reading-related N170 expertise to words - symbols Apple novel word forms. The pseudoword N170 differs from both the symbol N170 and the word N170, but the two A symbol E Figure 2 . Point-to-poin RP maps superimposed t differences (TAN wiOVA) betwe th Global Fiel ed Power n word and effects show distinct topographies. The pseudoword-symbol A. Point-to-point differences (TANOVA) between effect shows large differences at inferior (surrounding nega- word and symbol ERP maps superimposed with Glo- tive difference) and superior (central positive difference) bal Field Power. Significant differences (black bars) locations. The word-pseudoword effect is most pronounced between word and symbol processing were found in two over left occipito-temporal electrodes. Critical t-values are time windows corresponding to the N170 and P300 GFP 2.14 (p<0.05), 2.98 (p<0.01), and 4.14 (p<0.001). components. B. N170 maps for words and symbols and their difference t-map. Words elicited larger N170 nega- tivity than symbols at inferior occipito-temporal channels, especially over the left hemisphere. Note that the channels Topographic centroid analyses with the largest negative difference are located inferior to We used centroid measures (centers of gravity) of the pos- the classical 10-10 montage system. itive and negative fields on the scalp surface to character- ize the ERP topography [30-32]. The 3D locations of the positive and negative centroids were computed from all 129 electrode positions (in x-, y-, and z-Talairach space Global Field Power analysis [33]) weighted by their positive or negative values, respec- To assess overall map strength, we ran a repeated measure tively. Repeated measure ANOVAs were run on the cen- Analysis of Variance (ANOVA) on the N170 GFP value troid positions, separately for the overall reading-related separately for the overall reading-related contrast (word contrast (words vs. symbols), and the two contrasts vs. symbol), and for the two contrasts testing generaliza- testing for generalization to pseudowords (pseudowords tion of reading-related N170 specialization to novel word vs. symbols and words vs. pseudowords). Positive and forms (pseudowords vs. symbols and words vs. negative centroids were grouped in a factor "polarity", as pseudowords). positive and negative poles are often systematically related in ERP maps, and the three spatial coordinates There was no significant overall reading-related effect in were treated as multivariate dependent measures. Con- the N170 GFP measure, although map strength was some- trast main effects and polarity interactions are only what larger for symbols than for words (F(1,14) = 3.00, p reported if they differ significantly (p < 0.05) at the multi- = ns, see also Fig. 2). The additional comparisons revealed variate level. For multivariate significant effects, univariate that N170 GFP was larger in response to symbols than to tests were computed for the x-, y-, and z-axes, to character- pseudowords (F(1,14) = 6.07, p < 0.05), but did not differ ize the nature of the multivariate effect in 3D space. Con- between words and pseudowords (F(1,14) = 2.13, p = ns), trast main effects are referred to as "mean centroids" although map strength was slightly larger for words than (positive and negative centroids showed a similar pat- pseudowords. tern), and contrast-by-polarity interaction effects are Page 4 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 Table 2: Effects of the N170 topographic centroid analyses. Topographic effects (multivariate significant) x-axis y-axis z-axis Contrast (words vs. symbols) x polarity (positive vs. negative) p < 0.05 ns p < 0.001 Contrast (pseudowords vs. symbols) x polarity (positive vs. negative) ns* ns p < 0.001 Contrast (words vs. pseudowords) p < 0.05 ns Ns * non-significant trend (F(1,14) = 3.73, p < 0.1) mm yet superior and right-lateralized for symbols. The positive superior centroids showed a reversed pattern, as they were located more superior for words and more inferior for symbols (Fig. 4). inferior Clear topographic effects for the N170 responses to pseu- 30 dowords and symbols were also found in the centroid analysis, as indicated by different centroid distributions between pseudoword and symbol conditions (contrast x polarity F(3,12) = 7.10, p < 0.01). This difference was mainly found on the z-axis (F(1,14) = 20.94, p < 0.001, table 2). The negative centroids were located more inferior words for pseudowords and more superior for symbols, whereas pseudowords the positive centroids showed a reversed pattern (Fig. 4). symbols There was an additional non-significant trend on the x- 0 mm axis with the centroids more lateralized for symbols than -15 -10 -5 0 5 10 15 for pseudowords (F(1,14) = 3.74, p < 0.1). Positive an ord, and sym Figure 4 d n be ol top gativeo centr graphies oids of the N170 word, pseudow- Positive and negative centroids of the N170 word, Topographic effects between the N170 in response to pseudoword, and symbol topographies. Symbol centro- words and pseudowords were indicated by different mean ids show a different pattern from word and pseudoword centroid locations for words and pseudowords (F(3,12) = centroids, with a reversed polarity in the inferior-superior 3.59, p < 0.05). These effects appeared on the left-right x direction. In addition, the negative centroid is left-lateralized axis (F(1,14) = 4.73, p < 0.05, table 2). The mean centroids for words, but right-lateralized for symbols. The centroids were more left-lateralized for words than for pseudowords are also more left-lateralized for words than for pseudow- ords. Note that the centroids represent the ERP topography (Fig. 4). on the scalp surface and are by no means estimations of the underlying sources. Selected waveform analyses In order to allow comparisons with more conventional ERP analysis approaches, we also performed an analysis on left and right inferior occipito-temporal channel groups which have been shown to be most sensitive to referred to as "centroid distribution" (positive and nega- word-symbol differences [1]. Figure 1 illustrates the tive centroids showed a different pattern). specific channels included in the left and right groups, respectively. Repeated measure ANOVAs were run with As summarized in Table 2, the centroid analysis revealed the hemisphere factor (left vs. right channel group) clear overall reading-related effects in the N170, indicated separately for the overall reading-related contrast (words by a different centroid distribution between word and vs. symbols), and the two contrasts testing for generaliza- symbol maps (contrast x polarity, F(3,12) = 10.76, p < tion to pseudowords (pseudowords vs. symbols and 0.01). These differences appeared in both the analysis of words vs. pseudowords). the inferior-superior z coordinate axis (F(1,14) = 37.30, p < 0.001) and the analysis of left-right x coordinate axis A clear overall reading-related effect was seen in the (F(1,14) = 7.07, p < 0.05, table 2). As illustrated in Fig. 4, selected waveform analysis comparing word and symbol this interaction captures differences between negative cen- N170. The N170 amplitude was larger for words than for troids appearing as inferior and left-lateralized for words, Page 5 of 12 (page number not for citation purposes) left right Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 µV left occipito-temporal µV right occipito-temporal +1 +1 0 0 -1 -1 words pseudowords -2 -2 N170 N170 symbols -3 -3 -100 0 100 200 300 400 500 600 700 800 ms -100 0 100 200 300 400 500 600 700 800 ms Wa Figure 5 veforms at left and right inferior occipito-temporal channels Waveforms at left and right inferior occipito-temporal channels. The N170 is larger for words than for pseudowords and symbols, especially at the left hemisphere channels. Pseudowords have a larger N170 than symbols at both hemispheres, especially during the late part of the N170. symbols (contrast, F(1,14) = 9.93, p < 0.01), and this dif- word and symbol stimulus blocks demonstrated signifi- ference was larger over the left hemisphere (contrast x hem- cantly distinct topographies. The strongest topographic isphere, F(1,14) = 10.78, p < 0.01; Fig. 5). effect was found in a different centroid distribution on the inferior-superior coordinate axis. This effect reflected the The pseudoword-symbol contrast also revealed a signifi- occipito-temporal negativity and fronto-central positivity cant effect. Pseudowords elicited larger N170 amplitudes in the word maps and the occipito-parietal negativity and than symbols (F(1,14) = 7.18, p < 0.05; Fig. 5). Although fronto-polar positivity in the symbol maps. This differ- this difference was somewhat larger over the left hemi- ence also led to maximal effects in the t-map at inferior sphere, the interaction with hemisphere failed to reach sig- and superior electrodes. Indeed, the largest negative nificance (F(1,14) = 3.26, p < 0.1). effects were at the inferior edge of the channel array and might be missed by many conventional montages, which The word-pseudoword contrast also revealed a significant typically do not sample these regions [23,24,34]. An addi- effect in N170 amplitudes at inferior occipito-temporal tional topographic effect was found in a different laterali- channels. The amplitudes were larger for words than pseu- zation of the word and the symbol N170. The negative dowords (contrast, F(1,14) = 8.35, p < 0.05), which were centroids were left-lateralized for words, but right-lateral- more pronounced at the channels over the left than over ized for symbols. This lateralization difference was cor- the right hemisphere (contrast x hemisphere, F(1,14) = roborated in the selected waveform analysis, which 7.57, p < 0.05; Fig. 5). showed larger N170 amplitudes for words than symbols especially over the left inferior occipito-temporal channels. Discussion Summary of the results The present high-density ERP study clearly shows reading- The results also inform the generalization of reading- related expertise effects that occur early during processing related expertise in the N170 to novel word forms. The of visual words. Processing differences between words and pseudoword vs. symbols N170 contrast led to large t-val- novel symbol strings that control for basic visual features ues in the statistical difference map and demonstrated emerged in a time window corresponding to the N170 strong evidence for perceptual expertise elicited by novel ERP component. word forms. The topography of this N170 effect played out primarily as a shift in the inferior-superior dimension One of the central goals of this study involved extensive of the positive and negative centroids, very similar to the analysis of the topography of the N170 reading-related inferior-superior topographic N170 difference found perceptual expertise effect. The N170 responses elicited by between word and symbol centroids. In contrast, the Page 6 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 topographic effect in the left-right dimension only par- comparison in the present study also appeared in the tially generalized to pseudowords, showing just a non-sig- Zurich data, where it reached significance in the last two nificant trend for lateralization differences, which was thirds of the N170. Overall, the results suggest that word- corroborated in the selected waveform analysis. symbol differences in the N170 time range are robust markers for rapid specialization for reading, which can be Overall, N170 reading expertise effects do not appear to detected across different EEG systems and languages. fully generalize to pseudoword probes in English, as a generalization was found for the topographic inferior- Although the two studies showed similar topographic superior effect, but only partially for the topographic lat- N170 effects for the word-symbol comparison, reading- eralization effect. related specialization differed between the two languages when probed with novel word forms. For the The strongest evidence that generalization to novel word pseudoword-symbol comparison, both studies found sig- forms differs in lateralization came from the word-pseu- nificant differences in the inferior-superior dimension, doword comparison, in which the N170 centroids were similar to the one for the word-symbol comparison. In the more left-lateralized for words than for pseudowords. left-right dimension, however, the difference was signifi- This left-lateralization was corroborated in the additional cant in Zurich, but not in the present study. The strongest analysis at inferior occipito-temporal channel groups and evidence for a difference in generalization to in the statistical difference map. This demonstrates that pseudowords between the two studies was found in the the left-lateralized topographic effect of reading-related word-pseudoword comparison, where in contrast to the visual expertise in the N170 does not generalize well to significant lateralization difference in the present study, pseudowords. no difference at all was found in the Zurich study. This reflects the fact that the pseudoword topographies were Behavioral results, overall, served to ensure that the partic- left-lateralized for German speakers, but bilateral in the ipants demonstrated roughly equivalent levels of engage- present study. As we will discuss below, differences in ment with the different classes of stimuli, although subtle orthographic depth between the two languages may behavioral differences were revealed in the case of the explain this effect in particular and may shed some light symbol condition, in which slightly lower accuracy sug- on the characteristic left-lateralized topography of read- gests that detecting symbol strings was slightly more diffi- ing-related N170 specialization in general. cult compared to the other conditions. Interestingly, the equivalent speed and accuracy for target detection across Behavioral results were very similar in both studies. Repe- words and pseudowords demonstrates that the left-later- tition detection was high (>90%) in all conditions, with a alized word-pseudoword effect in the N170 is not slight advantage for detecting words and pseudowords dependent on processing differences assessed by behavio- compared to symbols in both studies. There was no differ- ral measures. ence in reaction time between conditions in any of the two studies. This shows that the difference in generaliza- Replication of effects across languages tion of reading-related expertise to pseudowords between The present study adopted a paradigm based on an earlier the two studies was not due to differences in overt behav- study in Zurich [1] that used a different EEG system with ioral responses. participants speaking a different language. The paradigm was identical in the two studies, except for language-spe- In addition to the generalization difference of reading- cific word and pseudoword stimuli, and the same analysis related N170 expertise to novel word forms, the two stud- strategy was used. This allows us to examine the two sets ies also differed in relative map strength of the N170 of results regarding similarities and differences. between the conditions. In the Zurich study, words and pseudowords had larger GFP than symbols, but in the Overall, the basic findings regarding reading-related spe- present study this relation tended to be reversed. Smaller cialization were successfully replicated, such as major overall amplitudes in the EEG for electrolyte-net systems word-symbol differences during the N170 and the P300 compared to electrogel-cap systems, as reported earlier time windows, and similar topographic N170 differences [35], should affect language and symbol stimuli equally. between words and symbols with larger negativities for Although the different word-symbol GFP ratios between words than for symbols at inferior occipito-temporal the studies could potentially result from differing symbol channels especially over the left hemisphere. Particularly, GFP, this seems unlikely because the two studies mainly the centroid analyses of the N170 maps showed the same differed with respect to the language stimuli. Thus, this robust differences between words and symbols in the infe- may suggest that words and pseudowords elicit relatively rior-superior dimension in the two studies. The left-later- smaller N170 amplitudes in English than in German alized topographic effect for the word-symbol which could be due to differences in orthographic depth Page 7 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 between the two languages. However, to exclude a possi- associated with different functional properties of reading- ble influence of the recording system, such language- related expertise. related N170 effects on GFP may be investigated in future studies using the same system for the two language The inferior-superior modulation of the N170 word-sym- groups. bol difference fully generalized to pseudowords in the present study. The same inferior-superior effect was also In contrast, reading-related N170 effects basically repli- found for the word-symbol and pseudoword-symbol cated their general topographies across languages, which contrast in the Zurich study with participants speaking a suggests that topography (rather than GFP) is a robust different language [1]. Since the inferior-superior topo- marker of reading-related specialization in the N170 graphic effect is robust across languages and generalizes to across EEG systems and language backgrounds. This, in novel word forms, it may reflect visual expertise for the turn, suggests that the topographic N170 differences familiarity of letters within strings. Visual expertise between the two studies found in the word-pseudoword reflected by the inferior-superior topographic N170 effect comparisons are valid markers for the influence of differ- may be related to expertise in other visual domains, such ent languages. as expertise for faces or objects [2,3]. The topographic centroid analyses not only revealed sim- Such speculation regarding various forms of expertise, ilar effects for reading-related N170 specialization in the however, requires additional investigation, as studies on word-symbol comparison between the two studies, the face or object expertise have not analyzed topographic centroid analysis also appeared as a more fruitful analysis effects beyond lateralization [4-6]. It remains to be tested strategy than the selected waveform analysis in the present whether the inferior-superior topographic effect in the study. Although the waveform analysis confirmed the lat- N170 relates to a functional property that is shared across eralization effects of the centroids, the pre-selected chan- different domains of visual expertise, or whether it repre- nel array prevented the detection of the strong sents a functional property characteristic of reading. topographic inferior-superior effects in the word-symbol and pseudoword-symbol N170 contrasts. This illustrates, Earlier findings suggested that the main characteristic fea- how the selection of particular channels in multi-channel ture of reading-related N170 specialization lies in its left- recordings could lead to biased results and conclusions. lateralized topography [1,3,8], contrasting the typically The centroid analysis method is a means for unbiased top- bilateral or right-lateralized N170 topographies for faces ographic ERP analyses, and has been proven useful to and objects of expertise [2,3]. The present results add detect topographic differences in earlier studies (e.g. [30- further evidence for this notion, but extend earlier find- 32]). ings, by showing that this left-lateralized topographic effect does not fully generalize to novel word forms in General Discussion English, as it did in German [1]. Thus, the left-lateralized The differences between word and symbol processing that topographic effect may be associated with functional appeared in the N170 component in the present study, properties that can be inferred from language differences corroborate findings from studies using MEG [10] or con- between English and German, especially with respect to ventional EEG systems [1,8,9,36,37] in supporting the pseudoword reading. general conclusion that processes in posterior brain regions that are activated within the first 200 msec are sen- English and German differ in the degree of orthographic sitive to reading-related experience. The present study depth, which is deep in English and shallow in German. extends earlier findings in two important ways: showing Orthographic depth refers to the level of consistency that the maximal effect of the negative difference appears whith which spelling maps onto word sounds (feedfor- at scalp locations that were not sampled previously, and ward consistency) and word sounds map onto spelling showing that reading-related expertise in the N170 has a (feedbackward consistency). In the case of pseudoword distinct functional organization in English which can be reading, the former plays an important role. Due to incon- detected when the expertise system is probed with sistency in spelling-to-sound mapping in English, the pro- pseudowords. nunciation of pseudowords is much more ambiguous in English than in German. Thus the left-lateralized topo- Reading-related expertise in the N170 appeared in two graphic N170 effect may specifically relate to processes topographic dimensions: the inferior-superior dimension involved in mapping letters onto word sounds. Mapping and the left-right dimension. These two topographic consistency has been demonstrated to be a central factor effects generalized to a different degree to pseudowords in modulating the rise of automaticity in information the present study, suggesting that the two effects may be processing [38]. The lack of a left-lateralization for English pseudowords may suggest that such processes are less Page 8 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 automatic in English [39], and are engaged to a lesser vation and N170 amplitude modulation for pseudoword degree while detecting pseudoword repetitions, because processing in English. repetition detection does not require explicit pronuncia- tion of the stimuli. Combined hemodynamic and electrophysiological stud- ies can also help localize the sources of the reading-related Such an interpretation also fits with the lack of N170 N170 specialization. Studies combining fMRI with MEG word-pseudoword differences in earlier studies in Finnish [41] and EEG [42] support the view that the word N170 and French [8,16]. Whereas Finnish orthography is shal- originates predominantely from inferior occipito-tempo- low, French orthography has some inconsistencies in ral regions, in agreement with source localization from sound-to-spelling mapping, but is rather consistent in studies using MEG and EEG alone [1,3,10]. The posterior spelling-to-sound mapping, which renders pseudoword left-lateralized effect for words in the current study is con- pronunciation less ambiguous in French [40]. sistent with sources in the left inferior occipito-temporal region, including the general region of the "Visual Word One study in English also did not find N170 differences Form Area", and suggests that such a left posterior region between words and pseudowords, but the participants demonstrates different patterns of neuronal responses to performed a lexical task that encouraged deeper language words vs. visual control stimuli within the first 200 msec processing of the pseudowords, which is in agreement of processing. with the automaticity hypothesis [21]. The notion of the Visual Word Form Area was first Another study in English found larger N170 amplitudes inspired by neuropsychological observations of "pure" for irregular pseudowords than for words, which may sug- alexia, or letter by letter reading, characterized by an ina- gest that the irregularity of the pseudowords led to an bility to read entire words, typically following damage to enhancement of the N170 similar to findings for conso- left-inferior-temporal regions with a maximal probability nant strings [14,15]. The effect of larger N170 amplitudes over fusiform gyrus (see [7] for review). Left fusiform for consonant strings compared to words may be related gyrus regions are also activated in metabolic studies con- to the well-replicated findings of larger N170 amplitudes trasting words and visual control stimuli, and this area has for low-frequency words than for high-frequency words been termed the Visual Word Form Area [7,42], suggesting [18-21]. For both the consonant strings and the word fre- a structure-function linkage between this region and early quency effects, no lateralization differences have been cognitive perceptual processes proposed in models of reported and the inferior-superior topographic effect has word recognition [38]. Although the left fusiform gyrus is not been investigated. Future studies may show whether typically activated in visual word tasks, this specialization this effect is related to the inferior-superior topographic does not necessarily exclude the participation of this effect in the present study, or whether it represents an region in other forms of processing, such as picture recog- additional reading-related modulation of the N170 nition, nor does it exclude the participation of additional component. regions in visual word processing (for review see [7,43]). Different levels of engagement in orthographic-to-phono- Previous neuroimaging studies of the putative "Visual logical processing might also explain the variable left-lat- Word Form Area" in the left fusiform gyrus have shown eralization of N170 related to reading in the literature sensitivity for orthographic regularity, with more activa- [3,12]. Thus, the degree of left-lateralization may vary tion for words and pseudowords than for nonwords (for according to language, task, and stimulus factors that a review see [7]). In contrast, sensitivity for familiarity of impact the degree to which visual, orthographic and pho- word forms was small (for a review see [7]), although a nological codes are engaged. more recent study suggests that activation may increase for words with low frequency and for pseudowords [44]. Evidence for language-specific effects on pseudoword In the present ERP study the N170 showed sensitivity for processing also comes from a PET study with English and the familiarity of word forms suggesting that this sensitiv- Italian subjects. During explicit and implicit pseudoword ity may be language-dependent, which might also apply processing, left posterior inferior temporal regions were for the fMRI results. However, there are additional reasons more activated in English subjects, whereas in Italian sub- that could explain different results between ERP and fMRI jects left superior temporal regions were more activated studies. Some N170 effects may be too transient to be cap- [25]. These results corroborate that pseudowords are proc- tured by the low temporal resolution of fMRI. Moreover, essed differently in languages that differ in consistency of it is also possible that the N170 effects for the word-pseu- spelling-to-sound mapping. However, future studies com- doword comparison do not originate from the left fusi- bining hemodynamic and electrophysiological methods form region, but from other left posterior regions are needed to clarify the relation between metabolic acti- contributing to the N170. Future research combining Page 9 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 fMRI and ERP in the same study, and examining factors German words were translated to English (high-frequency such as spelling-to-sound consistency patterns across lan- words in both languages), and the German pseudowords guages, and within words and pseudowords, may help to were replaced by regular English pseudowords. Both elucidate the nature of different neural contributions to words and pseudowords were printed with an initial cap- the N170 related to reading expertise. ital letter to match the visual characteristics of German nouns. Words, pseudowords, and symbol-strings were Conclusion matched for string length and contained 4.5 letters/sym- The present study provides further evidence that there are bols on average (range: 3–7), which also equaled the Ger- rapid perceptual processes in the brain that are specialized man string length. The experimental parameters were kept for reading. The N170 topography is a robust neurophys- identical to the Zurich study: stimuli were shown every iological marker for this specialization showing more 2050 ms for 700 ms in black on a white background 100 inferior and left-lateralized negativity for words compared cm away from the subject at a visual angle of 1.6–3.6 to symbols. The results extend this general finding via a degrees (shorter distance and smaller print size to keep the dense array and extended inferior coverage, demonstrat- same visual angle). In each condition, 72 stimuli were pre- ing that the maximal negative effect is more inferior than sented in 2 blocks containing 17% repetitions, which reported previously. Further characterization of the N170 served as targets. To keep the experiment context the same response in the current study suggests that reading-related as in the Zurich study, 2 blocks of pictures were also pre- perceptual expertise in the N170 can be characterized by sented within the same session, but these data are not at least two topographic effects which generalize to novel reported, as the different stimulus size and stimulus con- word forms to different degrees. The inferior-superior top- trast of the pictures would confound condition effects in ographic effect in the N170 fully generalized to novel the N170 component. In all blocks, participants were word forms, and may reflect expertise for letters or well- instructed to press a button with their right thumb when- ordered letter strings. Unlike in German, the left-lateral- ever they detected an immediate repetition. ized topographic effect in the N170 did not generalize to novel word forms in English. Inference from language dif- Electrophysiological Recording and Analyses ferences between English and German suggests that the The 129 channel ERPs were recorded using a geodesic sen- sor net [46] with a Cz reference. Data were sampled at 250 left-lateralized topographic effect in reading-related N170 specialization may reflect spelling-to-sound conversion, Hz/channel with filter settings 0.1–100 Hz and with cali- which might be less automaticly engaged in pseudoword brated technical zero baselines. Impedance was kept processing in English due to more ambiguous pronuncia- below 50 kΩ [47]. Using BESA software, channels with tion of novel word forms. excessive artifacts were spline interpolated (in average 3.5 channels per subject), and eye blinks were corrected (mul- tiple source eye correction method [48], as applied in the Methods Participants Zurich study). The data then were digitally bandpass fil- The data of 15 right-handed, native English speakers (19 tered (0.3–30 Hz), segmented (-150–850 ms), artifact to 29 years old) are presented. All subjects had normal or rejected (± 100 uV), and averaged according to non-target corrected-to-normal vision and their word reading and stimuli separately for the four conditions. Using Brain pseudoword decoding abilities [45] were within the nor- Vision Analyzer software, the averaged data were re-refer- mal range (within 2 SD of the norm mean). Although EEG enced to average reference ([27]), and highpass filtered (1 data of 20 subjects were obtained, data of 5 subjects were Hz) to further reduce slow wave drifts. After computing discarded due to low signal-to-noise ratios (3 subjects), Global Field Power (GFP) [27], grandmeans were com- bad net fit (1 subject), and outlier values in the ERP (>3 puted for all four stimulus conditions. SD, 1 subject). All subjects provided informed consent approved by the Weill-Cornell Institutional Review Board To assess differential processing of word and symbol Committee. strings, a Topographic Analysis of Variance (TANOVA, [26], part of the LORETA-Key software package, available Procedure at http://www.unizh.ch/keyinst on non-normalized (raw) To investigate rapid specialization for print, we used a par- ERP maps was computed for each time point. TANOVA adigm that was used in an earlier study in Zurich, Switzer- on raw maps detects all systematic amplitude differences land [1]. The main differences between the two studies are between two maps running a nonparametric randomiza- the different EEG systems (geodesic net vs. electrode caps) tion test [49] on the GFP of difference maps [26,27]. Note and the language of the stimuli and participants (English that differences resulting from TANOVA on raw maps can vs. German). The experiment used the same stimulus be due to different topographies, as well as due to differ- string conditions (words, pseudowords, symbol strings) ent map strengths. as [1], but while the symbol-strings were identical, the Page 10 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 For the N170 analysis, a time segment was selected References 1. Maurer U, Brem S, Bucher K, Brandeis D: Emerging neurophysio- between the two GFP minima before and after the N170 logical specialization for letter strings. J Cogn Neurosci in press. peak (144–248 ms) of the averaged word and symbol GFP 2. Bentin S, Allison T, Puce A, Perez E: Electrophysiological studies grandmeans, as in [1]. For the N170 segment maps, GFP of face perception in humans. J Cogn Neurosci 1996, 8(6):551-565. and 3D centroids were computed. GFP is the root mean 3. Rossion B, Joyce CA, Cottrell GW, Tarr MJ: Early lateralization square of the values at all electrodes. The positive 3D cen- and orientation tuning for face, word, and object processing in the visual cortex. Neuroimage 2003. troid is the voltage-weighted average of the locations of all 4. Tanaka JW, Curran T: A neural basis for expert object electrodes showing positive values; the negative 3D cen- recognition. Psychol Sci 2001, 12(1):43-47. troid is the analogous computed for electrodes with nega- 5. Gauthier I, Curran T, Curby KM, Collins D: Perceptual interfer- ence supports a non-modular account of face processing. Nat tive values. Centroid locations are shown in Talairach Neurosci 2003, 6(4):428-432. space [33]. GFP and centroid measures were analyzed in 6. Rossion B, Gauthier I, Goffaux V, Tarr MJ, Crommelinck M: Exper- repeated measure ANOVAs with a "contrast" factor (either tise training with novel objects leads to left-lateralized face- like electrophysiological responses. Psychol Sci 2002, words vs. symbols, pseudowords vs. symbols, or words vs. 13(3):250-257. pseudowords). For the centroid analyses "polarity" (posi- 7. McCandliss BD, Cohen L, Dehaene S: The visual word form area: expertise for reading in the fusiform gyrus. Trends Cogn Sci tive vs. negative) was included as an additional factor, and 2003, 7(7):293-299. the x-, y-, and z-axes were treated as multivariate depend- 8. Bentin S, Mouchetant-Rostaing Y, Giard MH, Echallier JF, Pernier J: ent measures. The selected waveform analysis in the N170 ERP manifestations of processing printed words at different psycholinguistic levels: time course and scalp distribution. J segment was computed with values from inferior occipito- Cogn Neurosci 1999, 11(3):235-260. temporal channel groups over the left (channels 57, 58, 9. Brem S, Lang-Dullenkopf A, Maurer U, Halder P, Bucher K, Brandeis 63, 64, 65, 69, and 70) and right (channels 90, 91, 95, 96, D: Neurophysiological signs of rapidly emerging visual exper- tise for symbol strings. Neuroreport 2005, 16(1):45-48. 97, 100, 101) hemisphere, as inferior occipito-temporal 10. Tarkiainen A, Helenius P, Hansen PC, Cornelissen PL, Salmelin R: regions were most sensitive to word-symbol differences in Dynamics of letter string perception in the human occipito- temporal cortex. Brain 1999, 122(Pt 11):2119-2132. earlier work [1]. This analysis was similar to the GFP anal- 11. Brandeis D, Lehmann D, Michel CM, Mingrone W: Mapping event- ysis with the additional "hemisphere" factor (left vs. related brain potential microstates to sentence endings. right). Brain Topogr 1995, 8(2):145-159. 12. Koenig T, Kochi K, Lehmann D: Event-related electric micro- states of the brain differ between words with visual and For the behavioral analysis two repeated measure ANO- abstract meaning. Electroencephalogr Clin Neurophysiol 1998, VAs were computed for accuracy and reaction time with 106(6):535-546. 13. Skrandies W: Evoked potential correlates of semantic mean- the "condition" factor (words vs. pseudowords vs. ing--A brain mapping study. Brain Res Cogn Brain Res 1998, symbols). 6(3):173-183. 14. McCandliss BD, Posner MI, Givon T: Brain plasticity in learning visual words. Cogn Psychol 1997, 33(1):88-110. List of abbreviations 15. Compton PE, Grossenbacher P, Posner MI, Tucker DM: A cogni- 3D: three-dimensional tive-anatomical approach to attention in lexical access. J Cogn Neurosci 1991, 3(4):304-312. 16. Wydell TN, Vuorinen T, Helenius P, Salmelin R: Neural correlates ANOVA: Analysis of Variance of letter-string length and lexicality during reading in a reg- ular orthography. J Cogn Neurosci 2003, 15(7):1052-1062. 17. Assadollahi R, Pulvermuller F: Early influences of word length EEG: Electroencephalography and frequency: a group study using MEG. Neuroreport 2003, 14(8):1183-1187. 18. Hauk O, Pulvermuller F: Effects of word length and frequency ERP: Event-related potential on the human event-related potential. Clin Neurophysiol 2004, 115(5):1090-1103. GFP: Global Field Power 19. Neville HJ, Mills DL, Lawson DS: Fractionating language: Differ- ent neural subsystems with different sensitive periods. Cereb Cortex 1992, 2:244-258. TANOVA: Topographic Analysis of Variance 20. Sereno SC, Brewer CC, O'Donnell PJ: Context effects in word recognition: evidence for early interactive processing. Psychol Sci 2003, 14(4):328-333. Authors' contributions 21. Sereno SC, Rayner K, Posner MI: Establishing a time-line of word UM, DB, and BDM designed the study and the data anal- recognition: evidence from eye movements and event- related potentials. Neuroreport 1998, 9(10):2195-2200. ysis strategy. UM and BDM shared the elaboration of the 22. Proverbio AM, Vecchi L, Zani A: From orthography to phonetics: paper. UM was responsible for data collection and data ERP measures of grapheme-to-phoneme conversion mecha- nisms in reading. J Cogn Neurosci 2004, 16(2):301-317. analysis. 23. Chatrian GE, Lettich E, Nelson PL: Ten percent electrode system for topographic studies of spontaneous and evoked EEG Acknowledgements activity. Am J EEG Technol 1985, 25:83-92. 24. Oostenveld R, Praamstra P: The five percent electrode system This research was supported by the Swiss National Science Foundation for high-resolution EEG and ERP measurements. Clin (Fellowship for Prospective Researchers: UM) and the US National Science Neurophysiol 2001, 112(4):713-719. Foundation (NSF 529112). We thank Sela Han for helping collecting the 25. Paulesu E, McCrory E, Fazio F, Menoncello L, Brunswick N, Cappa SF, data and Jason D. Zevin for helpful discussions. Cotelli M, Cossu G, Corte F, Lorusso M, Pesenti S, Gallagher A, Page 11 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 Perani D, Price C, Frith CD, Frith U: A cultural effect on brain 48. Berg P, Scherg M: A multiple source approach to the correc- function. Nat Neurosci 2000, 3(1):91-96. tion of eye artifacts. Electroencephalogr Clin Neurophysiol 1994, 26. Strik WK, Fallgatter AJ, Brandeis D, Pascual-Marqui RD: Three- 90(3):229-241. dimensional tomography of event-related potentials during 49. Holmes AP, Blair RC, Watson JD, Ford I: Nonparametric analysis response inhibition: evidence for phasic frontal lobe of statistic images from functional mapping experiments. J activation. Electroencephalogr Clin Neurophysiol 1998, Cereb Blood Flow Metab 1996, 16(1):7-22. 108(4):406-413. 27. Lehmann D, Skrandies W: Reference-free identification of com- ponents of checkerboard-evoked multichannel potential fields. Electroencephalogr Clin Neurophysiol 1980, 48(6):609-621. 28. Lehmann D, Skrandies W: Spatial analysis of evoked potentials in man--a review. Prog Neurobiol 1984, 23(3):227-250. 29. Brandeis D, Lehmann D: Event-related potentials of the brain and cognitive processes: approaches and applications. Neu- ropsychologia 1986, 24(1):151-168. 30. Brandeis D, Vitacco D, Steinhausen HC: Mapping brain electric micro-states in dyslexic children during reading. Acta Paedopsychiatr 1994, 56(3):239-247. 31. Maurer U, Bucher K, Brem S, Brandeis D: Development of the automatic mismatch response: from frontal positivity in kin- dergarten children to the mismatch negativity. Clin Neurophysiol 2003, 114(5):808-817. 32. Maurer U, Bucher K, Brem S, Brandeis D: Altered responses to tone and phoneme mismatch in kindergartners at familial dyslexia risk. Neuroreport 2003, 14(17):2245-2250. 33. Talairach J, Tournoux P: Co-planar Stereotaxic Atlas of the Human Brain. Stuttgart , Thieme; 1988. 34. Luu P, Ferree T: Determination of the Geodesic Sensor Nets' Average Electrode Positions and their 10-10 International Equivalents. In Technical Note Eugene, Oregon , Electrical Geodes- ics, Inc; 2000:1-11. 35. Kayser J, Tenke CE, Bhattacharya N, Stuart BK, Hudson J, Bruder GE: A direct comparison of Geodesic Sensor Net (128-channel) and conventional (30-channel) ERPs in tonal and phonetic oddball tasks. Psychophysiology 2000, 37:S17. 36. Schendan HE, Ganis G, Kutas M: Neurophysiological evidence for visual perceptual categorization of words and faces within 150 ms. Psychophysiology 1998, 35(3):240-251. 37. Zhang XL, Begleiter H, Porjesz B, Litke A: Visual object priming differs from visual word priming: an ERP study. Electroencepha- logr Clin Neurophysiol 1997, 102(3):200-215. 38. Schneider W, Shiffrin RM: Controlled and automatic human information processing: I. detection, search, and attention. Psychological Review 1977, 84:1-66. 39. Zevin JD, Balota DA: Priming and attentional control of lexical and sublexical pathways during naming. J Exp Psychol Learn Mem Cogn 2000, 26(1):121-135. 40. Ziegler JC, Perry C, Coltheart M: Speed of lexical and nonlexical processing in French: the case of the regularity effect. Psychon Bull Rev 2003, 10(4):947-953. 41. Dale AM, Liu AK, Fischl BR, Buckner RL, Belliveau JW, Lewine JD, Halgren E: Dynamic statistical parametric mapping: combin- ing fMRI and MEG for high-resolution imaging of cortical activity. Neuron 2000, 26(1):55-67. 42. Cohen L, Dehaene S, Naccache L, Lehericy S, Dehaene-Lambertz G, Henaff MA, Michel F: The visual word form area: spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients. Brain 2000, 123(Pt 2):291-307. 43. Price CJ, Devlin JT: The pro and cons of labelling a left occipi- totemporal region: "the visual word form area". Neuroimage Publish with Bio Med Central and every 2004, 22(1):477-479. scientist can read your work free of charge 44. Kronbichler M, Hutzler F, Wimmer H, Mair A, Staffen W, Ladurner G: The visual word form area and the frequency with which "BioMed Central will be the most significant development for words are encountered: evidence from a parametric fMRI disseminating the results of biomedical researc h in our lifetime." study. Neuroimage 2004, 21(3):946-953. Sir Paul Nurse, Cancer Research UK 45. Torgesen JK, Wagner RK, Rashotte CA: TOWRE - Test of word reading efficiency. Austin, Texas , PRO-ED; 1999. Your research papers will be: 46. Tucker DM: Spatial sampling of head electrical fields: the geo- available free of charge to the entire biomedical community desic sensor net. Electroencephalogr Clin Neurophysiol 1993, 87(3):154-163. peer reviewed and published immediately upon acceptance 47. Ferree TC, Luu P, Russell GS, Tucker DM: Scalp electrode imped- cited in PubMed and archived on PubMed Central ance, infection risk, and EEG data quality. Clin Neurophysiol 2001, 112(3):536-544. yours — you keep the copyright BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 12 of 12 (page number not for citation purposes) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavioral and Brain Functions Springer Journals

Fast, visual specialization for reading in English revealed by the topography of the N170 ERP response

Download PDF
12 pages

Loading next page...
 
/lp/springer-journals/fast-visual-specialization-for-reading-in-english-revealed-by-the-ShATdrO7De

References (55)

Publisher
Springer Journals
Copyright
Copyright © 2005 by Maurer et al; licensee BioMed Central Ltd.
Subject
Biomedicine; Neurosciences; Neurology; Behavioral Therapy; Psychiatry
eISSN
1744-9081
DOI
10.1186/1744-9081-1-13
pmid
16091138
Publisher site
See Article on Publisher Site

Abstract

Background: N170 effects associated with visual words may be related to perceptual expertise effects that have been demonstrated for faces and other extensively studied classes of visual stimuli. Although face and other object expertise effects are typically bilateral or right-lateralized, the spatial topography of reading-related N170 effects are often left-lateralized, providing potential insights into the unique aspects of reading-related perceptual expertise. Methods: Extending previous research in German [1], we use a high-density channel array to characterize the N170 topography for reading-related perceptual expertise in English, a language with inconsistent spelling-to-sound mapping. N170 effects related to overall reading-related expertise are defined by contrasting responses to visual words versus novel symbol strings. By contrasting each of these conditions to pseudowords, we examined how this reading-related N170 effect generalizes to well-ordered novel letter strings. Results: A sample-by-sample permutation test computed on word versus symbol ERP topographies revealed differences during two time windows corresponding to the N170 and P300 components. Topographic centroid analysis of the word and symbol N170 demonstrated significant differences in both left-right as well as inferior-superior dimensions. Words elicited larger N170 negativities than symbols at inferior occipito-temporal channels, with the maximal effect over left inferior regions often unsampled in conventional electrode montages. Further contrasts produced inferior-superior topographic effects for the pseudoword-symbol comparison and left-lateralized topographic effects for the word-pseudoword comparison. Conclusion: Fast specialized perception related to reading experience produces an N170 modulation detectable across different EEG systems and different languages. Characterization of such effects may be improved by sampling with greater spatial frequency recordings that sample inferior regions. Unlike in German, reading-related expertise effects in English produced only partial generalization in N170 responses to novel pseudowords. The topographic inferior-superior N170 differences may reflect general perceptual expertise for orthographic strings, as it was found for words and pseudowords across both languages. The topographic left-right N170 difference between words and pseudowords was only found in English, and may suggest that ambiguity in pronunciating novel pseudowords due to inconsistency in spelling-to-sound mapping influences early stages of letter string processing. Page 1 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 One recent study of native German speaking adults dem- Background The N170 is a component of the event-related potential onstrated sublexical N170 effects by contrasting responses (ERP) peaking between 150 and 200 ms and showing an to strings of novel symbols to letters strings grouped into occipito-temporally negative and fronto-centrally positive common orthographic patterns (familiar or novel word topography. It is strongly elicited by certain classes of vis- forms in German), demonstrating that specialization of ual stimuli, such as faces [2,3], relative to other visual con- processing as assessed by either lexical or sublexical trol stimuli. Investigations of the psychological principles strings is strongest at inferior occipito-temporal channels that drive the N170 to respond more strongly to some [1]. However, the region that demonstrated the peak effect classes of stimuli over others have demonstrated perceptual was at the edge of the montage of electrodes applied, sug- expertise effects across several classes of stimuli, including gesting that this effect might be better characterized by an enhanced N170 responses (relative to other object control electrode array, which covers regions inferior to the classi- stimuli) for bird experts viewing birds [4], car experts cal 10–20 or 10-10 electrode montages [23,24]. viewing cars [5], and has even been demonstrated for lab- oratory-induced expertise with 3D novel figures ("gree- The aim of the current study was to further characterize bles" [6]). These results support a potential relationship the nature of reading-related N170 expertise effects by between extensive visual experience with a stimulus applying a 129-channel array (geodesic sensor net, Electri- domain and alterations in visual processes within the first cal Geodesics, Inc.) that extends the coverage of the classi- 200 ms of perceptual identification. This framework of cal 10-10 montage [23,24] to more inferior regions, perceptual expertise may also account for experience- thereby providing adequate spatial sampling of the peak dependent changes in reading skill – a domain in which effect of interest (Fig. 1). This study adopts a paradigm extensive practice develops considerable visual expertise used in a previous N170 study conducted with German at the level of letter-strings and the pattern by which letters speaking subjects [1] to explore potential replicability of typically are combined to create visual word forms [7]. effects across languages that differ in the level of consist- ency of how letters map onto word sounds [25], as well as Neurophysiological studies have shown that skilled adult to examine the topography of the N170 with a greater spa- readers develop fast, perceptual identification processes tial sampling of inferior regions that might be critical to that are specialized for words and other letter strings, capturing effects produced near ventral posterior brain reflected by differences in N170 responses compared to regions (see [7] for review). EEG was recorded continu- control stimuli, such as symbol strings, that control for ously as participants actively monitored for an occasional visual features [1,8-10]. target, defined as an immediately repeated item (i.e. "one- back"), among a series of word, pseudoword, or symbol Unlike findings of right-lateralized or bilateral N170 string stimuli. Advantages of this paradigm include rela- responses for faces, N170 responses to word stimuli tively equal engagement for all classes of stimuli, which showed a left-lateralized topography [1,3,11-13]. Across can be assessed via behavioral responses to targets, as well studies, however, the degree of the left-lateralization var- the segregation of response-free trials from infrequent tar- ied between strong [3,12] and moderate [1,13]. get trials within ERP analyses. Some studies also showed that the N170 is sensitive for Results Behavior linguistic processing [14,15]. Consonant strings had larger N170 amplitudes than words [14,15], and sublexi- Participants detected targets with a high accuracy across cally irregular pseudowords were in between [14]. Other all conditions (>90% in each condition, see table 1), indi- studies, however, did not find N170 differences between cating that all conditions were relatively easy. Subtle con- words and pseudowords [1,8,16]. In one study, the differ- dition differences were detectable, however, via repeated ences between consonant strings and words were only measure ANOVA analyses, which revealed a main effect of found for lexical and semantic tasks, but not during stimulus condition for accuracy (F(2,13) = 6.23, p < 0.05), implicit reading [8], whereas it was found across seman- but not for reaction time (F(2,13) = 0.61, p = ns). Symbol tic, passive, and implicit viewing in another [14]. These strings were detected slightly less accurately than the other results suggest that N170 responses are somewhat variable two conditions. across experiments, which might be due to different task demands and presentation modes. Word-symbol differences in consecutive ERP maps To assess differential processing of words and symbol Word frequency effects in the N170 were more consist- strings over time, a Topographic Analysis of Variance ently found across studies, with low frequency words pro- (TANOVA, [26]) on non-normalized (raw) ERP maps was ducing more negative N170 amplitudes [17-21] (but see computed for each time point. TANOVA on raw maps also [22]). detects all systematic amplitude differences between two Page 2 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 Table 1: Behavioral results for detecting targets Words Pseudowords Symbols Accuracy (% correct) 98.4 95.5 90.1 Fp1 Fp2 Reaction time (ms) 609 599 542 F7 F8 F3 F4 Fz for the N170 component, we averaged samples across the time segment between the two GFP minima (based on the C3 C4 Cz T3/T7 T4/T8 average of word and symbol grandmeans) that marked the beginning and end of the N170 component as in [1]. P3 P4 Pz The N170 segment maps (144–248 ms) of the word, pseudoword, and symbol conditions were subsequently 57/TP9 57/TP10 T5/P7 T6/P8 analyzed to characterize GFP and topography effects across these stimulus groups. Topographic effects were O1 O2 tested using difference map t-statistics, centroid analyses, 65 91 and analyses of selected channels. First, overall reading- 69 70/PO9 90/PO10 95 related N170 effects were analyzed, comparing words vs. symbol strings. Second, generalization of reading-related N170 specialization to novel word forms was tested com- paring pseudowords both to symbols and to words. H Figure 1 igh-density 129-channel montage High-density 129-channel montage. Filled black dots Statistical difference map analyses indicate electrodes corresponding to the 10–20 system posi- The word N170 topography showed the largest negativity tions [34]. Inferior occipito-temporal channel groups used at occipito-temporal electrodes with a maximum over the for waveform illustration and additional lateralization analy- ses are marked with dotted circles. Note that the high-den- left hemisphere and the largest positivity at fronto-central sity montage extends both the 10-10 and 10-5 montages [24] electrodes. The symbol N170 topography showed the for an additional inferior row (approx. 5% of the Nasion- largest negativity at occipito-parietal electrodes with a Inion distance). maximum over the right hemisphere and the largest posi- tivity at fronto-polar electrodes. The two topographies clearly differed, as shown by the large t-values in the sta- tistical difference map (Fig. 2). The maximal effects were found at parietal electrodes and at left inferior temporal and occipito-temporal electrodes at the edge of the chan- nel array (Fig. 2). maps (i.e. including all 129 electrodes). Accordingly, word and symbol processing differed (p < 0.01, to adjust N170 responses to pseudowords and symbols clearly dif- for multiple comparisons) during two separate time win- fered in the statistical difference map (Fig. 3). The maxi- dows, from 160–244 ms and from 324–512 ms, largely mal effects were found at parietal electrodes and overlapping with the Global Field Power (GFP [27]) peaks bilaterally at inferior electrodes at the edge of the elec- of the N170 and P300 ERP components (Fig. 2). trode array. N170 time window The statistical difference map also indicated clear N170 Our approach to analyzing topographic effects in the effects between responses to words and pseudowords (Fig. N170 time window follows an ERP mapping approach 3). These effects were left-lateralized, as they were found [28,29], designed to take full advantage of information at many occipito-temporal channels over the left hemi- from all the channels in the high-density channel array. sphere, but hardly over the right hemisphere. The pseu- According to this approach, ERPs are seen as a series of doword N170 topography resembled the word N170, but maps changing in Global Field Power (GFP [27]) and showed a more bilateral negativity and a positivity cen- topography over time. Moreover, ERP topographies tend tered around the fronto-central midline electrodes, in to remain stable for short periods of time, typically chang- contrast to the more left-lateralized negativity and positiv- ing at time points with low GFP. To get a robust measure ity of the word N170 (Fig. 3). Page 3 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 TANOVA (word-symbol difference) GFP words p<0.01 symbols N170 P300 p<0.05 0 -100 0 100 200 300 400 500 600 700 800ms symbol N170 t-map word N170 Gener wo Figure 3 rd fa olization of reading rms -related N170 expertise to novel -2 +2 0µV +6 -6 0t Generalization of reading-related N170 expertise to words - symbols Apple novel word forms. The pseudoword N170 differs from both the symbol N170 and the word N170, but the two A symbol E Figure 2 . Point-to-poin RP maps superimposed t differences (TAN wiOVA) betwe th Global Fiel ed Power n word and effects show distinct topographies. The pseudoword-symbol A. Point-to-point differences (TANOVA) between effect shows large differences at inferior (surrounding nega- word and symbol ERP maps superimposed with Glo- tive difference) and superior (central positive difference) bal Field Power. Significant differences (black bars) locations. The word-pseudoword effect is most pronounced between word and symbol processing were found in two over left occipito-temporal electrodes. Critical t-values are time windows corresponding to the N170 and P300 GFP 2.14 (p<0.05), 2.98 (p<0.01), and 4.14 (p<0.001). components. B. N170 maps for words and symbols and their difference t-map. Words elicited larger N170 nega- tivity than symbols at inferior occipito-temporal channels, especially over the left hemisphere. Note that the channels Topographic centroid analyses with the largest negative difference are located inferior to We used centroid measures (centers of gravity) of the pos- the classical 10-10 montage system. itive and negative fields on the scalp surface to character- ize the ERP topography [30-32]. The 3D locations of the positive and negative centroids were computed from all 129 electrode positions (in x-, y-, and z-Talairach space Global Field Power analysis [33]) weighted by their positive or negative values, respec- To assess overall map strength, we ran a repeated measure tively. Repeated measure ANOVAs were run on the cen- Analysis of Variance (ANOVA) on the N170 GFP value troid positions, separately for the overall reading-related separately for the overall reading-related contrast (word contrast (words vs. symbols), and the two contrasts vs. symbol), and for the two contrasts testing generaliza- testing for generalization to pseudowords (pseudowords tion of reading-related N170 specialization to novel word vs. symbols and words vs. pseudowords). Positive and forms (pseudowords vs. symbols and words vs. negative centroids were grouped in a factor "polarity", as pseudowords). positive and negative poles are often systematically related in ERP maps, and the three spatial coordinates There was no significant overall reading-related effect in were treated as multivariate dependent measures. Con- the N170 GFP measure, although map strength was some- trast main effects and polarity interactions are only what larger for symbols than for words (F(1,14) = 3.00, p reported if they differ significantly (p < 0.05) at the multi- = ns, see also Fig. 2). The additional comparisons revealed variate level. For multivariate significant effects, univariate that N170 GFP was larger in response to symbols than to tests were computed for the x-, y-, and z-axes, to character- pseudowords (F(1,14) = 6.07, p < 0.05), but did not differ ize the nature of the multivariate effect in 3D space. Con- between words and pseudowords (F(1,14) = 2.13, p = ns), trast main effects are referred to as "mean centroids" although map strength was slightly larger for words than (positive and negative centroids showed a similar pat- pseudowords. tern), and contrast-by-polarity interaction effects are Page 4 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 Table 2: Effects of the N170 topographic centroid analyses. Topographic effects (multivariate significant) x-axis y-axis z-axis Contrast (words vs. symbols) x polarity (positive vs. negative) p < 0.05 ns p < 0.001 Contrast (pseudowords vs. symbols) x polarity (positive vs. negative) ns* ns p < 0.001 Contrast (words vs. pseudowords) p < 0.05 ns Ns * non-significant trend (F(1,14) = 3.73, p < 0.1) mm yet superior and right-lateralized for symbols. The positive superior centroids showed a reversed pattern, as they were located more superior for words and more inferior for symbols (Fig. 4). inferior Clear topographic effects for the N170 responses to pseu- 30 dowords and symbols were also found in the centroid analysis, as indicated by different centroid distributions between pseudoword and symbol conditions (contrast x polarity F(3,12) = 7.10, p < 0.01). This difference was mainly found on the z-axis (F(1,14) = 20.94, p < 0.001, table 2). The negative centroids were located more inferior words for pseudowords and more superior for symbols, whereas pseudowords the positive centroids showed a reversed pattern (Fig. 4). symbols There was an additional non-significant trend on the x- 0 mm axis with the centroids more lateralized for symbols than -15 -10 -5 0 5 10 15 for pseudowords (F(1,14) = 3.74, p < 0.1). Positive an ord, and sym Figure 4 d n be ol top gativeo centr graphies oids of the N170 word, pseudow- Positive and negative centroids of the N170 word, Topographic effects between the N170 in response to pseudoword, and symbol topographies. Symbol centro- words and pseudowords were indicated by different mean ids show a different pattern from word and pseudoword centroid locations for words and pseudowords (F(3,12) = centroids, with a reversed polarity in the inferior-superior 3.59, p < 0.05). These effects appeared on the left-right x direction. In addition, the negative centroid is left-lateralized axis (F(1,14) = 4.73, p < 0.05, table 2). The mean centroids for words, but right-lateralized for symbols. The centroids were more left-lateralized for words than for pseudowords are also more left-lateralized for words than for pseudow- ords. Note that the centroids represent the ERP topography (Fig. 4). on the scalp surface and are by no means estimations of the underlying sources. Selected waveform analyses In order to allow comparisons with more conventional ERP analysis approaches, we also performed an analysis on left and right inferior occipito-temporal channel groups which have been shown to be most sensitive to referred to as "centroid distribution" (positive and nega- word-symbol differences [1]. Figure 1 illustrates the tive centroids showed a different pattern). specific channels included in the left and right groups, respectively. Repeated measure ANOVAs were run with As summarized in Table 2, the centroid analysis revealed the hemisphere factor (left vs. right channel group) clear overall reading-related effects in the N170, indicated separately for the overall reading-related contrast (words by a different centroid distribution between word and vs. symbols), and the two contrasts testing for generaliza- symbol maps (contrast x polarity, F(3,12) = 10.76, p < tion to pseudowords (pseudowords vs. symbols and 0.01). These differences appeared in both the analysis of words vs. pseudowords). the inferior-superior z coordinate axis (F(1,14) = 37.30, p < 0.001) and the analysis of left-right x coordinate axis A clear overall reading-related effect was seen in the (F(1,14) = 7.07, p < 0.05, table 2). As illustrated in Fig. 4, selected waveform analysis comparing word and symbol this interaction captures differences between negative cen- N170. The N170 amplitude was larger for words than for troids appearing as inferior and left-lateralized for words, Page 5 of 12 (page number not for citation purposes) left right Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 µV left occipito-temporal µV right occipito-temporal +1 +1 0 0 -1 -1 words pseudowords -2 -2 N170 N170 symbols -3 -3 -100 0 100 200 300 400 500 600 700 800 ms -100 0 100 200 300 400 500 600 700 800 ms Wa Figure 5 veforms at left and right inferior occipito-temporal channels Waveforms at left and right inferior occipito-temporal channels. The N170 is larger for words than for pseudowords and symbols, especially at the left hemisphere channels. Pseudowords have a larger N170 than symbols at both hemispheres, especially during the late part of the N170. symbols (contrast, F(1,14) = 9.93, p < 0.01), and this dif- word and symbol stimulus blocks demonstrated signifi- ference was larger over the left hemisphere (contrast x hem- cantly distinct topographies. The strongest topographic isphere, F(1,14) = 10.78, p < 0.01; Fig. 5). effect was found in a different centroid distribution on the inferior-superior coordinate axis. This effect reflected the The pseudoword-symbol contrast also revealed a signifi- occipito-temporal negativity and fronto-central positivity cant effect. Pseudowords elicited larger N170 amplitudes in the word maps and the occipito-parietal negativity and than symbols (F(1,14) = 7.18, p < 0.05; Fig. 5). Although fronto-polar positivity in the symbol maps. This differ- this difference was somewhat larger over the left hemi- ence also led to maximal effects in the t-map at inferior sphere, the interaction with hemisphere failed to reach sig- and superior electrodes. Indeed, the largest negative nificance (F(1,14) = 3.26, p < 0.1). effects were at the inferior edge of the channel array and might be missed by many conventional montages, which The word-pseudoword contrast also revealed a significant typically do not sample these regions [23,24,34]. An addi- effect in N170 amplitudes at inferior occipito-temporal tional topographic effect was found in a different laterali- channels. The amplitudes were larger for words than pseu- zation of the word and the symbol N170. The negative dowords (contrast, F(1,14) = 8.35, p < 0.05), which were centroids were left-lateralized for words, but right-lateral- more pronounced at the channels over the left than over ized for symbols. This lateralization difference was cor- the right hemisphere (contrast x hemisphere, F(1,14) = roborated in the selected waveform analysis, which 7.57, p < 0.05; Fig. 5). showed larger N170 amplitudes for words than symbols especially over the left inferior occipito-temporal channels. Discussion Summary of the results The present high-density ERP study clearly shows reading- The results also inform the generalization of reading- related expertise effects that occur early during processing related expertise in the N170 to novel word forms. The of visual words. Processing differences between words and pseudoword vs. symbols N170 contrast led to large t-val- novel symbol strings that control for basic visual features ues in the statistical difference map and demonstrated emerged in a time window corresponding to the N170 strong evidence for perceptual expertise elicited by novel ERP component. word forms. The topography of this N170 effect played out primarily as a shift in the inferior-superior dimension One of the central goals of this study involved extensive of the positive and negative centroids, very similar to the analysis of the topography of the N170 reading-related inferior-superior topographic N170 difference found perceptual expertise effect. The N170 responses elicited by between word and symbol centroids. In contrast, the Page 6 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 topographic effect in the left-right dimension only par- comparison in the present study also appeared in the tially generalized to pseudowords, showing just a non-sig- Zurich data, where it reached significance in the last two nificant trend for lateralization differences, which was thirds of the N170. Overall, the results suggest that word- corroborated in the selected waveform analysis. symbol differences in the N170 time range are robust markers for rapid specialization for reading, which can be Overall, N170 reading expertise effects do not appear to detected across different EEG systems and languages. fully generalize to pseudoword probes in English, as a generalization was found for the topographic inferior- Although the two studies showed similar topographic superior effect, but only partially for the topographic lat- N170 effects for the word-symbol comparison, reading- eralization effect. related specialization differed between the two languages when probed with novel word forms. For the The strongest evidence that generalization to novel word pseudoword-symbol comparison, both studies found sig- forms differs in lateralization came from the word-pseu- nificant differences in the inferior-superior dimension, doword comparison, in which the N170 centroids were similar to the one for the word-symbol comparison. In the more left-lateralized for words than for pseudowords. left-right dimension, however, the difference was signifi- This left-lateralization was corroborated in the additional cant in Zurich, but not in the present study. The strongest analysis at inferior occipito-temporal channel groups and evidence for a difference in generalization to in the statistical difference map. This demonstrates that pseudowords between the two studies was found in the the left-lateralized topographic effect of reading-related word-pseudoword comparison, where in contrast to the visual expertise in the N170 does not generalize well to significant lateralization difference in the present study, pseudowords. no difference at all was found in the Zurich study. This reflects the fact that the pseudoword topographies were Behavioral results, overall, served to ensure that the partic- left-lateralized for German speakers, but bilateral in the ipants demonstrated roughly equivalent levels of engage- present study. As we will discuss below, differences in ment with the different classes of stimuli, although subtle orthographic depth between the two languages may behavioral differences were revealed in the case of the explain this effect in particular and may shed some light symbol condition, in which slightly lower accuracy sug- on the characteristic left-lateralized topography of read- gests that detecting symbol strings was slightly more diffi- ing-related N170 specialization in general. cult compared to the other conditions. Interestingly, the equivalent speed and accuracy for target detection across Behavioral results were very similar in both studies. Repe- words and pseudowords demonstrates that the left-later- tition detection was high (>90%) in all conditions, with a alized word-pseudoword effect in the N170 is not slight advantage for detecting words and pseudowords dependent on processing differences assessed by behavio- compared to symbols in both studies. There was no differ- ral measures. ence in reaction time between conditions in any of the two studies. This shows that the difference in generaliza- Replication of effects across languages tion of reading-related expertise to pseudowords between The present study adopted a paradigm based on an earlier the two studies was not due to differences in overt behav- study in Zurich [1] that used a different EEG system with ioral responses. participants speaking a different language. The paradigm was identical in the two studies, except for language-spe- In addition to the generalization difference of reading- cific word and pseudoword stimuli, and the same analysis related N170 expertise to novel word forms, the two stud- strategy was used. This allows us to examine the two sets ies also differed in relative map strength of the N170 of results regarding similarities and differences. between the conditions. In the Zurich study, words and pseudowords had larger GFP than symbols, but in the Overall, the basic findings regarding reading-related spe- present study this relation tended to be reversed. Smaller cialization were successfully replicated, such as major overall amplitudes in the EEG for electrolyte-net systems word-symbol differences during the N170 and the P300 compared to electrogel-cap systems, as reported earlier time windows, and similar topographic N170 differences [35], should affect language and symbol stimuli equally. between words and symbols with larger negativities for Although the different word-symbol GFP ratios between words than for symbols at inferior occipito-temporal the studies could potentially result from differing symbol channels especially over the left hemisphere. Particularly, GFP, this seems unlikely because the two studies mainly the centroid analyses of the N170 maps showed the same differed with respect to the language stimuli. Thus, this robust differences between words and symbols in the infe- may suggest that words and pseudowords elicit relatively rior-superior dimension in the two studies. The left-later- smaller N170 amplitudes in English than in German alized topographic effect for the word-symbol which could be due to differences in orthographic depth Page 7 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 between the two languages. However, to exclude a possi- associated with different functional properties of reading- ble influence of the recording system, such language- related expertise. related N170 effects on GFP may be investigated in future studies using the same system for the two language The inferior-superior modulation of the N170 word-sym- groups. bol difference fully generalized to pseudowords in the present study. The same inferior-superior effect was also In contrast, reading-related N170 effects basically repli- found for the word-symbol and pseudoword-symbol cated their general topographies across languages, which contrast in the Zurich study with participants speaking a suggests that topography (rather than GFP) is a robust different language [1]. Since the inferior-superior topo- marker of reading-related specialization in the N170 graphic effect is robust across languages and generalizes to across EEG systems and language backgrounds. This, in novel word forms, it may reflect visual expertise for the turn, suggests that the topographic N170 differences familiarity of letters within strings. Visual expertise between the two studies found in the word-pseudoword reflected by the inferior-superior topographic N170 effect comparisons are valid markers for the influence of differ- may be related to expertise in other visual domains, such ent languages. as expertise for faces or objects [2,3]. The topographic centroid analyses not only revealed sim- Such speculation regarding various forms of expertise, ilar effects for reading-related N170 specialization in the however, requires additional investigation, as studies on word-symbol comparison between the two studies, the face or object expertise have not analyzed topographic centroid analysis also appeared as a more fruitful analysis effects beyond lateralization [4-6]. It remains to be tested strategy than the selected waveform analysis in the present whether the inferior-superior topographic effect in the study. Although the waveform analysis confirmed the lat- N170 relates to a functional property that is shared across eralization effects of the centroids, the pre-selected chan- different domains of visual expertise, or whether it repre- nel array prevented the detection of the strong sents a functional property characteristic of reading. topographic inferior-superior effects in the word-symbol and pseudoword-symbol N170 contrasts. This illustrates, Earlier findings suggested that the main characteristic fea- how the selection of particular channels in multi-channel ture of reading-related N170 specialization lies in its left- recordings could lead to biased results and conclusions. lateralized topography [1,3,8], contrasting the typically The centroid analysis method is a means for unbiased top- bilateral or right-lateralized N170 topographies for faces ographic ERP analyses, and has been proven useful to and objects of expertise [2,3]. The present results add detect topographic differences in earlier studies (e.g. [30- further evidence for this notion, but extend earlier find- 32]). ings, by showing that this left-lateralized topographic effect does not fully generalize to novel word forms in General Discussion English, as it did in German [1]. Thus, the left-lateralized The differences between word and symbol processing that topographic effect may be associated with functional appeared in the N170 component in the present study, properties that can be inferred from language differences corroborate findings from studies using MEG [10] or con- between English and German, especially with respect to ventional EEG systems [1,8,9,36,37] in supporting the pseudoword reading. general conclusion that processes in posterior brain regions that are activated within the first 200 msec are sen- English and German differ in the degree of orthographic sitive to reading-related experience. The present study depth, which is deep in English and shallow in German. extends earlier findings in two important ways: showing Orthographic depth refers to the level of consistency that the maximal effect of the negative difference appears whith which spelling maps onto word sounds (feedfor- at scalp locations that were not sampled previously, and ward consistency) and word sounds map onto spelling showing that reading-related expertise in the N170 has a (feedbackward consistency). In the case of pseudoword distinct functional organization in English which can be reading, the former plays an important role. Due to incon- detected when the expertise system is probed with sistency in spelling-to-sound mapping in English, the pro- pseudowords. nunciation of pseudowords is much more ambiguous in English than in German. Thus the left-lateralized topo- Reading-related expertise in the N170 appeared in two graphic N170 effect may specifically relate to processes topographic dimensions: the inferior-superior dimension involved in mapping letters onto word sounds. Mapping and the left-right dimension. These two topographic consistency has been demonstrated to be a central factor effects generalized to a different degree to pseudowords in modulating the rise of automaticity in information the present study, suggesting that the two effects may be processing [38]. The lack of a left-lateralization for English pseudowords may suggest that such processes are less Page 8 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 automatic in English [39], and are engaged to a lesser vation and N170 amplitude modulation for pseudoword degree while detecting pseudoword repetitions, because processing in English. repetition detection does not require explicit pronuncia- tion of the stimuli. Combined hemodynamic and electrophysiological stud- ies can also help localize the sources of the reading-related Such an interpretation also fits with the lack of N170 N170 specialization. Studies combining fMRI with MEG word-pseudoword differences in earlier studies in Finnish [41] and EEG [42] support the view that the word N170 and French [8,16]. Whereas Finnish orthography is shal- originates predominantely from inferior occipito-tempo- low, French orthography has some inconsistencies in ral regions, in agreement with source localization from sound-to-spelling mapping, but is rather consistent in studies using MEG and EEG alone [1,3,10]. The posterior spelling-to-sound mapping, which renders pseudoword left-lateralized effect for words in the current study is con- pronunciation less ambiguous in French [40]. sistent with sources in the left inferior occipito-temporal region, including the general region of the "Visual Word One study in English also did not find N170 differences Form Area", and suggests that such a left posterior region between words and pseudowords, but the participants demonstrates different patterns of neuronal responses to performed a lexical task that encouraged deeper language words vs. visual control stimuli within the first 200 msec processing of the pseudowords, which is in agreement of processing. with the automaticity hypothesis [21]. The notion of the Visual Word Form Area was first Another study in English found larger N170 amplitudes inspired by neuropsychological observations of "pure" for irregular pseudowords than for words, which may sug- alexia, or letter by letter reading, characterized by an ina- gest that the irregularity of the pseudowords led to an bility to read entire words, typically following damage to enhancement of the N170 similar to findings for conso- left-inferior-temporal regions with a maximal probability nant strings [14,15]. The effect of larger N170 amplitudes over fusiform gyrus (see [7] for review). Left fusiform for consonant strings compared to words may be related gyrus regions are also activated in metabolic studies con- to the well-replicated findings of larger N170 amplitudes trasting words and visual control stimuli, and this area has for low-frequency words than for high-frequency words been termed the Visual Word Form Area [7,42], suggesting [18-21]. For both the consonant strings and the word fre- a structure-function linkage between this region and early quency effects, no lateralization differences have been cognitive perceptual processes proposed in models of reported and the inferior-superior topographic effect has word recognition [38]. Although the left fusiform gyrus is not been investigated. Future studies may show whether typically activated in visual word tasks, this specialization this effect is related to the inferior-superior topographic does not necessarily exclude the participation of this effect in the present study, or whether it represents an region in other forms of processing, such as picture recog- additional reading-related modulation of the N170 nition, nor does it exclude the participation of additional component. regions in visual word processing (for review see [7,43]). Different levels of engagement in orthographic-to-phono- Previous neuroimaging studies of the putative "Visual logical processing might also explain the variable left-lat- Word Form Area" in the left fusiform gyrus have shown eralization of N170 related to reading in the literature sensitivity for orthographic regularity, with more activa- [3,12]. Thus, the degree of left-lateralization may vary tion for words and pseudowords than for nonwords (for according to language, task, and stimulus factors that a review see [7]). In contrast, sensitivity for familiarity of impact the degree to which visual, orthographic and pho- word forms was small (for a review see [7]), although a nological codes are engaged. more recent study suggests that activation may increase for words with low frequency and for pseudowords [44]. Evidence for language-specific effects on pseudoword In the present ERP study the N170 showed sensitivity for processing also comes from a PET study with English and the familiarity of word forms suggesting that this sensitiv- Italian subjects. During explicit and implicit pseudoword ity may be language-dependent, which might also apply processing, left posterior inferior temporal regions were for the fMRI results. However, there are additional reasons more activated in English subjects, whereas in Italian sub- that could explain different results between ERP and fMRI jects left superior temporal regions were more activated studies. Some N170 effects may be too transient to be cap- [25]. These results corroborate that pseudowords are proc- tured by the low temporal resolution of fMRI. Moreover, essed differently in languages that differ in consistency of it is also possible that the N170 effects for the word-pseu- spelling-to-sound mapping. However, future studies com- doword comparison do not originate from the left fusi- bining hemodynamic and electrophysiological methods form region, but from other left posterior regions are needed to clarify the relation between metabolic acti- contributing to the N170. Future research combining Page 9 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 fMRI and ERP in the same study, and examining factors German words were translated to English (high-frequency such as spelling-to-sound consistency patterns across lan- words in both languages), and the German pseudowords guages, and within words and pseudowords, may help to were replaced by regular English pseudowords. Both elucidate the nature of different neural contributions to words and pseudowords were printed with an initial cap- the N170 related to reading expertise. ital letter to match the visual characteristics of German nouns. Words, pseudowords, and symbol-strings were Conclusion matched for string length and contained 4.5 letters/sym- The present study provides further evidence that there are bols on average (range: 3–7), which also equaled the Ger- rapid perceptual processes in the brain that are specialized man string length. The experimental parameters were kept for reading. The N170 topography is a robust neurophys- identical to the Zurich study: stimuli were shown every iological marker for this specialization showing more 2050 ms for 700 ms in black on a white background 100 inferior and left-lateralized negativity for words compared cm away from the subject at a visual angle of 1.6–3.6 to symbols. The results extend this general finding via a degrees (shorter distance and smaller print size to keep the dense array and extended inferior coverage, demonstrat- same visual angle). In each condition, 72 stimuli were pre- ing that the maximal negative effect is more inferior than sented in 2 blocks containing 17% repetitions, which reported previously. Further characterization of the N170 served as targets. To keep the experiment context the same response in the current study suggests that reading-related as in the Zurich study, 2 blocks of pictures were also pre- perceptual expertise in the N170 can be characterized by sented within the same session, but these data are not at least two topographic effects which generalize to novel reported, as the different stimulus size and stimulus con- word forms to different degrees. The inferior-superior top- trast of the pictures would confound condition effects in ographic effect in the N170 fully generalized to novel the N170 component. In all blocks, participants were word forms, and may reflect expertise for letters or well- instructed to press a button with their right thumb when- ordered letter strings. Unlike in German, the left-lateral- ever they detected an immediate repetition. ized topographic effect in the N170 did not generalize to novel word forms in English. Inference from language dif- Electrophysiological Recording and Analyses ferences between English and German suggests that the The 129 channel ERPs were recorded using a geodesic sen- sor net [46] with a Cz reference. Data were sampled at 250 left-lateralized topographic effect in reading-related N170 specialization may reflect spelling-to-sound conversion, Hz/channel with filter settings 0.1–100 Hz and with cali- which might be less automaticly engaged in pseudoword brated technical zero baselines. Impedance was kept processing in English due to more ambiguous pronuncia- below 50 kΩ [47]. Using BESA software, channels with tion of novel word forms. excessive artifacts were spline interpolated (in average 3.5 channels per subject), and eye blinks were corrected (mul- tiple source eye correction method [48], as applied in the Methods Participants Zurich study). The data then were digitally bandpass fil- The data of 15 right-handed, native English speakers (19 tered (0.3–30 Hz), segmented (-150–850 ms), artifact to 29 years old) are presented. All subjects had normal or rejected (± 100 uV), and averaged according to non-target corrected-to-normal vision and their word reading and stimuli separately for the four conditions. Using Brain pseudoword decoding abilities [45] were within the nor- Vision Analyzer software, the averaged data were re-refer- mal range (within 2 SD of the norm mean). Although EEG enced to average reference ([27]), and highpass filtered (1 data of 20 subjects were obtained, data of 5 subjects were Hz) to further reduce slow wave drifts. After computing discarded due to low signal-to-noise ratios (3 subjects), Global Field Power (GFP) [27], grandmeans were com- bad net fit (1 subject), and outlier values in the ERP (>3 puted for all four stimulus conditions. SD, 1 subject). All subjects provided informed consent approved by the Weill-Cornell Institutional Review Board To assess differential processing of word and symbol Committee. strings, a Topographic Analysis of Variance (TANOVA, [26], part of the LORETA-Key software package, available Procedure at http://www.unizh.ch/keyinst on non-normalized (raw) To investigate rapid specialization for print, we used a par- ERP maps was computed for each time point. TANOVA adigm that was used in an earlier study in Zurich, Switzer- on raw maps detects all systematic amplitude differences land [1]. The main differences between the two studies are between two maps running a nonparametric randomiza- the different EEG systems (geodesic net vs. electrode caps) tion test [49] on the GFP of difference maps [26,27]. Note and the language of the stimuli and participants (English that differences resulting from TANOVA on raw maps can vs. German). The experiment used the same stimulus be due to different topographies, as well as due to differ- string conditions (words, pseudowords, symbol strings) ent map strengths. as [1], but while the symbol-strings were identical, the Page 10 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 For the N170 analysis, a time segment was selected References 1. Maurer U, Brem S, Bucher K, Brandeis D: Emerging neurophysio- between the two GFP minima before and after the N170 logical specialization for letter strings. J Cogn Neurosci in press. peak (144–248 ms) of the averaged word and symbol GFP 2. Bentin S, Allison T, Puce A, Perez E: Electrophysiological studies grandmeans, as in [1]. For the N170 segment maps, GFP of face perception in humans. J Cogn Neurosci 1996, 8(6):551-565. and 3D centroids were computed. GFP is the root mean 3. Rossion B, Joyce CA, Cottrell GW, Tarr MJ: Early lateralization square of the values at all electrodes. The positive 3D cen- and orientation tuning for face, word, and object processing in the visual cortex. Neuroimage 2003. troid is the voltage-weighted average of the locations of all 4. Tanaka JW, Curran T: A neural basis for expert object electrodes showing positive values; the negative 3D cen- recognition. Psychol Sci 2001, 12(1):43-47. troid is the analogous computed for electrodes with nega- 5. Gauthier I, Curran T, Curby KM, Collins D: Perceptual interfer- ence supports a non-modular account of face processing. Nat tive values. Centroid locations are shown in Talairach Neurosci 2003, 6(4):428-432. space [33]. GFP and centroid measures were analyzed in 6. Rossion B, Gauthier I, Goffaux V, Tarr MJ, Crommelinck M: Exper- repeated measure ANOVAs with a "contrast" factor (either tise training with novel objects leads to left-lateralized face- like electrophysiological responses. Psychol Sci 2002, words vs. symbols, pseudowords vs. symbols, or words vs. 13(3):250-257. pseudowords). For the centroid analyses "polarity" (posi- 7. McCandliss BD, Cohen L, Dehaene S: The visual word form area: expertise for reading in the fusiform gyrus. Trends Cogn Sci tive vs. negative) was included as an additional factor, and 2003, 7(7):293-299. the x-, y-, and z-axes were treated as multivariate depend- 8. Bentin S, Mouchetant-Rostaing Y, Giard MH, Echallier JF, Pernier J: ent measures. The selected waveform analysis in the N170 ERP manifestations of processing printed words at different psycholinguistic levels: time course and scalp distribution. J segment was computed with values from inferior occipito- Cogn Neurosci 1999, 11(3):235-260. temporal channel groups over the left (channels 57, 58, 9. Brem S, Lang-Dullenkopf A, Maurer U, Halder P, Bucher K, Brandeis 63, 64, 65, 69, and 70) and right (channels 90, 91, 95, 96, D: Neurophysiological signs of rapidly emerging visual exper- tise for symbol strings. Neuroreport 2005, 16(1):45-48. 97, 100, 101) hemisphere, as inferior occipito-temporal 10. Tarkiainen A, Helenius P, Hansen PC, Cornelissen PL, Salmelin R: regions were most sensitive to word-symbol differences in Dynamics of letter string perception in the human occipito- temporal cortex. Brain 1999, 122(Pt 11):2119-2132. earlier work [1]. This analysis was similar to the GFP anal- 11. Brandeis D, Lehmann D, Michel CM, Mingrone W: Mapping event- ysis with the additional "hemisphere" factor (left vs. related brain potential microstates to sentence endings. right). Brain Topogr 1995, 8(2):145-159. 12. Koenig T, Kochi K, Lehmann D: Event-related electric micro- states of the brain differ between words with visual and For the behavioral analysis two repeated measure ANO- abstract meaning. Electroencephalogr Clin Neurophysiol 1998, VAs were computed for accuracy and reaction time with 106(6):535-546. 13. Skrandies W: Evoked potential correlates of semantic mean- the "condition" factor (words vs. pseudowords vs. ing--A brain mapping study. Brain Res Cogn Brain Res 1998, symbols). 6(3):173-183. 14. McCandliss BD, Posner MI, Givon T: Brain plasticity in learning visual words. Cogn Psychol 1997, 33(1):88-110. List of abbreviations 15. Compton PE, Grossenbacher P, Posner MI, Tucker DM: A cogni- 3D: three-dimensional tive-anatomical approach to attention in lexical access. J Cogn Neurosci 1991, 3(4):304-312. 16. Wydell TN, Vuorinen T, Helenius P, Salmelin R: Neural correlates ANOVA: Analysis of Variance of letter-string length and lexicality during reading in a reg- ular orthography. J Cogn Neurosci 2003, 15(7):1052-1062. 17. Assadollahi R, Pulvermuller F: Early influences of word length EEG: Electroencephalography and frequency: a group study using MEG. Neuroreport 2003, 14(8):1183-1187. 18. Hauk O, Pulvermuller F: Effects of word length and frequency ERP: Event-related potential on the human event-related potential. Clin Neurophysiol 2004, 115(5):1090-1103. GFP: Global Field Power 19. Neville HJ, Mills DL, Lawson DS: Fractionating language: Differ- ent neural subsystems with different sensitive periods. Cereb Cortex 1992, 2:244-258. TANOVA: Topographic Analysis of Variance 20. Sereno SC, Brewer CC, O'Donnell PJ: Context effects in word recognition: evidence for early interactive processing. Psychol Sci 2003, 14(4):328-333. Authors' contributions 21. Sereno SC, Rayner K, Posner MI: Establishing a time-line of word UM, DB, and BDM designed the study and the data anal- recognition: evidence from eye movements and event- related potentials. Neuroreport 1998, 9(10):2195-2200. ysis strategy. UM and BDM shared the elaboration of the 22. Proverbio AM, Vecchi L, Zani A: From orthography to phonetics: paper. UM was responsible for data collection and data ERP measures of grapheme-to-phoneme conversion mecha- nisms in reading. J Cogn Neurosci 2004, 16(2):301-317. analysis. 23. Chatrian GE, Lettich E, Nelson PL: Ten percent electrode system for topographic studies of spontaneous and evoked EEG Acknowledgements activity. Am J EEG Technol 1985, 25:83-92. 24. Oostenveld R, Praamstra P: The five percent electrode system This research was supported by the Swiss National Science Foundation for high-resolution EEG and ERP measurements. Clin (Fellowship for Prospective Researchers: UM) and the US National Science Neurophysiol 2001, 112(4):713-719. Foundation (NSF 529112). We thank Sela Han for helping collecting the 25. Paulesu E, McCrory E, Fazio F, Menoncello L, Brunswick N, Cappa SF, data and Jason D. Zevin for helpful discussions. Cotelli M, Cossu G, Corte F, Lorusso M, Pesenti S, Gallagher A, Page 11 of 12 (page number not for citation purposes) Behavioral and Brain Functions 2005, 1:13 http://www.behavioralandbrainfunctions.com/content/1/1/13 Perani D, Price C, Frith CD, Frith U: A cultural effect on brain 48. Berg P, Scherg M: A multiple source approach to the correc- function. Nat Neurosci 2000, 3(1):91-96. tion of eye artifacts. Electroencephalogr Clin Neurophysiol 1994, 26. Strik WK, Fallgatter AJ, Brandeis D, Pascual-Marqui RD: Three- 90(3):229-241. dimensional tomography of event-related potentials during 49. Holmes AP, Blair RC, Watson JD, Ford I: Nonparametric analysis response inhibition: evidence for phasic frontal lobe of statistic images from functional mapping experiments. J activation. Electroencephalogr Clin Neurophysiol 1998, Cereb Blood Flow Metab 1996, 16(1):7-22. 108(4):406-413. 27. Lehmann D, Skrandies W: Reference-free identification of com- ponents of checkerboard-evoked multichannel potential fields. Electroencephalogr Clin Neurophysiol 1980, 48(6):609-621. 28. Lehmann D, Skrandies W: Spatial analysis of evoked potentials in man--a review. Prog Neurobiol 1984, 23(3):227-250. 29. Brandeis D, Lehmann D: Event-related potentials of the brain and cognitive processes: approaches and applications. Neu- ropsychologia 1986, 24(1):151-168. 30. Brandeis D, Vitacco D, Steinhausen HC: Mapping brain electric micro-states in dyslexic children during reading. Acta Paedopsychiatr 1994, 56(3):239-247. 31. Maurer U, Bucher K, Brem S, Brandeis D: Development of the automatic mismatch response: from frontal positivity in kin- dergarten children to the mismatch negativity. Clin Neurophysiol 2003, 114(5):808-817. 32. Maurer U, Bucher K, Brem S, Brandeis D: Altered responses to tone and phoneme mismatch in kindergartners at familial dyslexia risk. Neuroreport 2003, 14(17):2245-2250. 33. Talairach J, Tournoux P: Co-planar Stereotaxic Atlas of the Human Brain. Stuttgart , Thieme; 1988. 34. Luu P, Ferree T: Determination of the Geodesic Sensor Nets' Average Electrode Positions and their 10-10 International Equivalents. In Technical Note Eugene, Oregon , Electrical Geodes- ics, Inc; 2000:1-11. 35. Kayser J, Tenke CE, Bhattacharya N, Stuart BK, Hudson J, Bruder GE: A direct comparison of Geodesic Sensor Net (128-channel) and conventional (30-channel) ERPs in tonal and phonetic oddball tasks. Psychophysiology 2000, 37:S17. 36. Schendan HE, Ganis G, Kutas M: Neurophysiological evidence for visual perceptual categorization of words and faces within 150 ms. Psychophysiology 1998, 35(3):240-251. 37. Zhang XL, Begleiter H, Porjesz B, Litke A: Visual object priming differs from visual word priming: an ERP study. Electroencepha- logr Clin Neurophysiol 1997, 102(3):200-215. 38. Schneider W, Shiffrin RM: Controlled and automatic human information processing: I. detection, search, and attention. Psychological Review 1977, 84:1-66. 39. Zevin JD, Balota DA: Priming and attentional control of lexical and sublexical pathways during naming. J Exp Psychol Learn Mem Cogn 2000, 26(1):121-135. 40. Ziegler JC, Perry C, Coltheart M: Speed of lexical and nonlexical processing in French: the case of the regularity effect. Psychon Bull Rev 2003, 10(4):947-953. 41. Dale AM, Liu AK, Fischl BR, Buckner RL, Belliveau JW, Lewine JD, Halgren E: Dynamic statistical parametric mapping: combin- ing fMRI and MEG for high-resolution imaging of cortical activity. Neuron 2000, 26(1):55-67. 42. Cohen L, Dehaene S, Naccache L, Lehericy S, Dehaene-Lambertz G, Henaff MA, Michel F: The visual word form area: spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients. Brain 2000, 123(Pt 2):291-307. 43. Price CJ, Devlin JT: The pro and cons of labelling a left occipi- totemporal region: "the visual word form area". Neuroimage Publish with Bio Med Central and every 2004, 22(1):477-479. scientist can read your work free of charge 44. Kronbichler M, Hutzler F, Wimmer H, Mair A, Staffen W, Ladurner G: The visual word form area and the frequency with which "BioMed Central will be the most significant development for words are encountered: evidence from a parametric fMRI disseminating the results of biomedical researc h in our lifetime." study. Neuroimage 2004, 21(3):946-953. Sir Paul Nurse, Cancer Research UK 45. Torgesen JK, Wagner RK, Rashotte CA: TOWRE - Test of word reading efficiency. Austin, Texas , PRO-ED; 1999. Your research papers will be: 46. Tucker DM: Spatial sampling of head electrical fields: the geo- available free of charge to the entire biomedical community desic sensor net. Electroencephalogr Clin Neurophysiol 1993, 87(3):154-163. peer reviewed and published immediately upon acceptance 47. Ferree TC, Luu P, Russell GS, Tucker DM: Scalp electrode imped- cited in PubMed and archived on PubMed Central ance, infection risk, and EEG data quality. Clin Neurophysiol 2001, 112(3):536-544. yours — you keep the copyright BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 12 of 12 (page number not for citation purposes)

Journal

Behavioral and Brain Functions – Springer Journals

Published: Aug 9, 2005

There are no references for this article.