Increasing expertise to a novel script modulates the visual N1 ERP in healthy adults

Abstract

Neural tuning to print develops when children learn to read and is reflected by a more pronounced left occipito-temporal negativity to orthographic stimuli as compared to non-orthographic false fonts or symbols after around 150–250 ms in their N1, a visual event-related potential (ERP). In adults, initial expertise for a novel script can emerge in less than 2 hours through repeated exposure or training. Here, we aimed to assess changes in the visual N1 related to the process of learning associations between unknown written characters and familiar, spoken syllables or words. Thirty-two healthy literate adults learned to associate a set of foreign characters with either syllables or German words within a single experimental session. EEG was recorded during a visual one-back character repetition detection task in which trained characters, untrained characters and familiar letters were presented before and after the training. A bilateral occipito-temporal increase in the N1 negativity with training was only found for the newly learned characters, but not for the control characters. In conclusion, the present data indicate that expertise to novel characters can be induced by a short character–sound association training and is reflected by a bilateral modulation of the visual N1 amplitude. However, no differentiation was found regarding the comparison of word or syllable training, indicating that the visual N1 most likely reflects gaining expertise driven by phonological associations common to both training types.

Keywords

Learning to link print and speech forms the basis of learning to read in all languages. During the process of learning print–speech correspondences, the brain is tuned to effectively process the newly learned characters (Brem et al., 2010). While in all writing systems the correspondences between characters and speech units have to be learned, major differences exist concerning the unit size of speech sound information associated with single characters (Ziegler & Goswami, 2005). This unit size ranges from single speech sounds in transparent alphabetic writing systems to syllables and morphemes (whole words) in syllabic, morphosyllabic and logographic writing systems. Accordingly, in literates a single character can denote either solely phonological information or a combination of phonological and lexical information.

One of the earliest signs for neural print tuning and visual expertise in the ERP of the electroencephalogram (EEG) is the visual occipito-temporal N1 (N170) reflecting activity of the occipito-temporal cortex after around 150–250 ms, as seen in intracranial recordings (Allison, McCarthy, Nobre, Puce, & Belger, 1994; Nobre, Allison, & McCarthy, 1994), functional magnetic resonance imaging studies (Cohen et al., 2000; Vinckier et al., 2007) or source estimations (Brem et al., 2006, 2010; Maurer, Brem, Bucher, & Brandeis, 2005). An enhanced visual N1 has been linked to perceptual expertise in essential domains of everyday life, such as face, letter, or word processing (Bentin, Mouchetant-Rostaing, Giard, Echallier, & Pernier, 1999; Brem et al., 2005; Curran, Tanaka, & Weiskopf, 2002; Maurer et al., 2005; Rossion, Joyce, Cottrell, & Tarr, 2003; Wong, Gauthier, Woroch, DeBuse, & Curran, 2005) but is also seen in individuals who are highly specialized in particular domains such as birds, dogs (Tanaka & Curran, 2001), fingerprints (Busey & Vanderkolk, 2005), cars (Gauthier, Curran, Curby, & Collins, 2003) or computer-generated novel objects (Rossion, Gauthier, Goffaux, Tarr, & Crommelinck, 2002). The typical left occipito-temporal maximum of the N1 for print clearly differs from the right lateralized or bilateral negativity seen for faces, objects or numbers (Park, Chiang, Brannon, & Woldorff, 2014; Rossion et al., 2003). Regarding visual print expertise, varying levels of neural tuning and sensitivity have recently been described for the visual N1. A coarse level of visual expertise to print is reflected by differential processing of words and objects or non-orthographic symbol strings (Bentin et al., 1999; Brem et al., 2006; Maurer et al., 2005; Schendan, Ganis, & Kutas, 1998). However, a somewhat more fine-tuned system is required for differentiating between print and closely matched false font or pseudofont character strings (Brem et al., 2013; Wong et al., 2005). Finally, a high level of expertise is necessary to differentiate between words and consonant strings (Proverbio, Vecchi, & Zani, 2004; Zhao et al., 2014).

Developmental ERP studies have clarified that the different levels of visual expertise to print emerge with reading acquisition in childhood, and together with functional magnetic resonance imaging (fMRI) findings suggest a phonologically-guided tuning of ventral-occipito-temporal regions to print (Brem et al., 2010; Sandak et al., 2004; Schlaggar & McCandliss, 2007). While illiterate kindergarten children do not show a print-sensitive N1, a more pronounced left occipito-temporal negativity in the visual N1 to words and orthographic stimuli as compared to non-orthographic false fonts or symbols after around 150–250 ms is already initialized after basic grapheme-phoneme correspondence training in kindergarteners (Brem et al., 2010). This print sensitive response is established in regular school children 1–2 years after starting formal reading instruction at school (Brem et al., 2013; Eberhard-Moscicka, Jost, Raith, & Maurer, 2015; Maurer et al., 2006; Zhao et al., 2014) and also depends on literacy skills in children (Araújo, Bramão, Faísca, Petersson, & Reis, 2012; Fraga González et al., 2014) and adults (Pegado et al., 2014). Furthermore, the print sensitivity of the visual N1 or its magnetoencephalographic analogue in adults is not limited to the first, native writing system (Maurer, Zevin, & McCandliss, 2008) nor is it specific to alphabetic writing systems. In non-alphabetic writing systems such as Korean, Japanese or Chinese (Kim & Kim, 2006; Maurer et al., 2008; Shirahama, Ohta, Takashima, Matsushima, & Okubo, 2004; Wong et al., 2005), similar N1 expertise effects in adults and children have been reported (Cao, Li, Zhao, Lin, & Weng, 2011). It has recently been shown that coarse and even fine levels of expertise to print can develop quickly and within 1 year of reading instruction, depending on previous experience (Zhao et al., 2014).

Studies with adults indicate that a certain level of expertise for a novel script, reflected by an increase in the N1 amplitude, can emerge in less than 2h through repeated exposure (Brem et al., 2005) or specific training (Maurer, Blau, Yoncheva, & McCandliss, 2010; Yoncheva, Blau, Maurer, & McCandliss, 2010; Yoncheva, Wise, & McCandliss, 2015). Moreover, the characteristic lateralization of the maximal N1 negativity over left occipito-temporal sites after training seems to depend on the literacy level (Pegado et al., 2014) and task requirements such as an attentional focus on learning character–phoneme rather than whole-word associations during reading instruction (Maurer et al., 2010; Yoncheva et al., 2010, 2015). In accordance with ERP results, training character–phoneme associations also induced changes in the activity of occipito-temporal regions in fMRI studies of adults (Hashimoto & Sakai, 2004; Xue, Chen, Jin, & Dong, 2006) and children (Brem et al., 2010; James, 2010). Despite considerable research on visual expertise to print in recent years, it remains unclear whether the neural N1 print tuning effect is mainly driven by lexical processing, phonological processing, or mere expertise and familiarity to a specific writing system. To date, most studies have focused on how whole words of a novel writing system are processed after specific trainings in alphabetic and non-alphabetic writing systems (Maurer et al., 2010; McCandliss, Posner, & Givon, 1997; Xue et al., 2006; Yoncheva et al., 2010, 2016), whereas it has hardly been examined how single characters are processed when associated with either speech sounds or lexical information.

Therefore, we compare the short-term effects of two different training types on the visual N1 during implicit processing of single characters. One training focused on learning solely phonological (character–syllable) and the other on lexico-phonological (character–word) associations to clarify the impact of lexical and phonological processes on neural tuning and lateralization of the visual N1. An implicit visual character processing task before and after a short training of either character–sound or character–word associations was used to examine the training effects in two matched groups of healthy adults.

Methods

Groups and participants

In this study, 32 healthy, right-handed and native German speaking adults (9 male) aged between 20.8 and 33.9 years participated, and completed the experiment which included a training session nested within two EEG recordings and a behavioral follow-up session. None of the participants reported a history of neurological or psychiatric disorders (aside from one participant who reported to have attention-deficit hyperactivity disorder in childhood) and none had any knowledge of a foreign writing system (Japanese, Chinese, Indian fonts).

The 32 participants were assigned to two groups (see Table 1) matched for age, sex, word, pseudoword (SLRT-II; Moll & Landerl, 2010) and sentence (SLS; Auer, Gruber, Mayringer, & Wimmer, 2005) reading scores (for all p > .15) and learned to associate 12 Japanese or Chinese characters with either spoken syllables (syllable training group ST) or whole words (word training group WT).

Table 1.

Demographic data, behavioural data on reading competence, training and EEG task performance.

	Syllable–training group (ST)	Word–training group (WT)	t test	Effect size
Variables	M (SD)	M (SD)	p	Cohen’s d
Behavioural and demographic measures
Trained script Japanese:Chinese (n)	10:6	8:8
Sex (female:male)	13:3	10:6
Age (years)	26.6 (3.2)	25.8 (3.7)	.526	−0.231
Interval EEG-Follow-up (days)	7.6 (1.6)	7.5 (1.3)	.810	−0.069
Training duration (min)	50.9 (13.0)	39.3 (7.3)	.004	−1.1
Word reading (SLRT-II) fluency^a	129.0 (16.0)	124.0 (14.0)	.348	−0.333
Pseudoword reading (SLRT-II) fluency^a	83.0 (19.0)	82.0 (15.0)	.777	−0.058
Sentence reading (SLS)^b	57.0 (10.0)	53.0 (10.0)	.178	−0.4
EEG session performance
Training (% correct)	96.1 (5.0)	99.6 (11.0)	.013	0.41
EEG Task, Pre hits (%)	92.3 (8.7)	91.1 (6.9)	.675	−0.153
EEG Task, Pre false-alarms (%)	4.7 (3.6)	4.4 (2.5)	.780	−0.097
EEG Task, Pre omissions (%)	7.7 (8.7)	5.6 (6.9)	.675	−0.267
EEG Task, Pre RT (ms)	523.3 (82.2)	521.3 (69.9)	.165	−0.026
EEG Task, Post hits (%)	90.2 (7.6)	85.3 (11.6)	.722	−0.5
EEG Task, Post false-alarms (%)	3.9 (2.0)	3.1 (1.8)	.258	−0.42
EEG Task, Post omissions (%)	9.8 (7.6)	14.7 (11.6)	.165	0.5
EEG Task, Post RT (ms)	530.3 (74.2)	545.2 (87.6)	.609	0.184
Performance in follow-up session
Training performance follow-up (% correct)	89.1 (0.8)	94.2 (7.3)	.133	0.982
Discrimination Task 1 hits (% correct)	94.8 (7.4)	94.8 (8.5)	1.000	0
Discrimination Task 1 time (s)	31.7 (20.1)	31.4 (12.0)	.966	−0.015
Discrimination Task 2 hits (% correct)	97.4 (4.0)	96.3 (7.4)	.625	−0.185
Discrimination Task 2 time (s)	89.3 (74.2)	80.9 (31.9)	.681	−0.147

Note. ^aNumber of correctly read items per minute.

^bNumber of correctly read sentences in three minutes.

All participants received a present (5–40 CHF) for their participation and provided written informed consent to the study approved by the local ethics commission.

The study was divided into two sessions. In the first session, the EEG was recorded before and after training character–syllable or character–word associations. The same visual and audiovisual tasks were performed before and after the computerized training of foreign characters. Here, we focus on the results of the implicit visual character processing task.

Approximately 1 week (mean 8 ± 1 days, range: 6–12 days) later, a follow-up session of behavioral assessments was conducted. The test battery included speeded word, pseudoword and sentence reading tests (Auer, Gruber, Mayringer, & Wimmer, 2005; Moll & Landerl, 2010) to assess the reading skills in the participants’ native language. To examine sustained training effects, the participants were additionally tested for the learned characters using two versions of a visual character recognition and discrimination task. In the first task, participants had to mark as quickly as possible those 12 characters in a list of 48 characters (novel and learned Chinese and Japanese intermixed) that they had learned. In the second, simpler task, participants again marked the 12 learned items, but now in a list of 24 items (containing only characters of the learned font, but novel and learned characters intermixed). The accuracy and time to complete these tasks were recorded. At the end of the follow-up assessment, the same test level of the computerized training game used to derive learning performance at the end of the training session was repeated.

Syllable and word training

The training was conducted with the Graphogame software, a computerized training game developed at the University of Jyväskylä in Finland and adjusted for the use of the present study by our group (Karipidis et al., 2017; Lyytinen, Erskine, Kujala, Ojanen, & Richardson, 2009; Lyytinen, Ronimus, Alanko, Poikkeus, & Taanila, 2007; Saine, Lerkkanen, Ahonen, Tolvanen, & Lyytinen, 2011). Four versions of the training were programmed in which either Japanese or Chinese characters were paired with either spoken syllables (nonsense syllables, e.g., “te,” “ma,” “nu”) or spoken words (all concrete, German one-syllable nouns e.g., “tier” [animal], “mond” [moon], “nuss” [nut]). The participants training Japanese characters did not encounter any Chinese characters during their computerized training and vice versa. The participants were given as much time as they needed to learn the complete set of 12 characters (cf. Figure 2). After half of the training session was completed, participants took a 10-min break. The training was adaptive and divided into different levels, including instruction levels where new characters were introduced, training levels where the correspondences of sounds and characters were practiced and support levels in which items with a high error rate were repeated. Overall, the training included 17 different levels and a test level at the end of the training to derive a measure of the participants’ learning performance.

Figure 1.

Implicit one-back repetition detection task.

Figure 2.

Conditions of the implicit one-back detection task.

During the training, the participants were instructed to choose the correct character among distractors for a presented speech sound. If the chosen character was correct, the character was highlighted by a green frame, whereas if the chosen character was wrong, the wrong character disappeared and the subject had to click on the remaining correct character. The training was adaptive by increasing the number of distractors (max. 5 distractors) when the subject performed well and decreasing them (min. 1 distractor) when errors were made. In the test level at the end of the training and the follow-up assessment, no feedback was given to the participants and every character appeared five times, always together with five distractors.

Implicit visual one-back character processing task

The implicit visual one-back repetition detection task included five different conditions (cf. Figure 1), each of which included 12 different single characters: familiar letters (L: Arial font), unfamiliar false font (FF) characters, Japanese (Hiragana) and Chinese characters/symbols that were either trained (TC) or served as untrained control characters (UC), and novel fonts (N). The N condition comprised two different Indian fonts, each of which was only used once, either before or after training to control for potential repetition and visual familiarity effects in the ERPs (cf. Figure 2). The 12 chosen characters of the two Indian fonts (Gurmuhi alphabet, Karmic, or Oriya alphabet, Utkal) were similar in their visual complexity and their order of appearance (pre-/post-training) was counterbalanced across participants. The participants were instructed to respond by a left mouse click as quickly as possible whenever two identical characters appeared on the screen consecutively. The experiment included 60 (20%) target stimuli (immediate character repetitions), whereby each character was repeated once as a target. Each character was shown in black on a white background five times throughout the experiment, always followed by a centered fixation cross to minimize eye movements. All 300 stimuli and targets were presented for 100 ms each in randomized order with jittered interstimulus intervals (mean = 1700 ms, range: 1450–1950 ms, 100 ms steps) to reduce preparatory potentials. Of note, three participants of the ST group performed the implicit visual one-back detection task without the N condition but with otherwise equal task design: these participants are thus still included in the main analyses (see statistics).

EEG recording and analysis

For the EEG recordings, the participants sat in a small chamber 1.33 m away from a 19-inch computer screen. Stimuli were presented with Presentation v. 13.1 (Neurobehavioral Systems, Inc) in font size 56. The EEG was recorded with a sampling rate of 500 Hz using a QUICKAMP 72 channel (Brain Products GmbH) amplifier with an average reference. 65 EEG channels (one bipolar, PO8 against POz) were recorded. AFz was used as ground. The electrodes were placed according to the international 10-20 system, except for Fp1’/Fp2’ and O1’/O2’, which were placed more laterally for a more even coverage of the scalp. Additional electrodes were placed on the following positions: Fpz, FCz, CPz, POz, Oz, Iz, AF1/2, F5/6, FC1/2, FC3/4, FC5/6, FT7/8, FT9/10, C1/2, C5/6, CP1/2, CP3/4, CP5/6, TP7/8, TP9/10, P5/6, PO1/2, PO7/8, PO9/10 and OI1/2. Two ocular electrodes were placed 1 cm lateral and below the outer canthus of the eyes (LE, RE). For all electrodes, impedances were kept below 20kΩ.

EEG data were processed with Vision Analyzer (Versions 1.56, and 2.1, BrainProducts GmbH, Munich, Germany). The data were filtered (0.1 to 30 Hz, 50 Hz notch) and artifact intervals exceeding ± 300 µV in all channels (except for channels LE/RE, Fp1’/Fp2’ close to the eyes an artifact criterion of ± 1000 µV was used) were excluded from further processing before correcting the artifacts dominated by blinks, lateral or vertical eye movements with an independent component analysis (ICA) approach. A joint average reference was computed for all EEG channels, including the bipolar recorded channel (PO8). Remaining artifacts in the corrected EEG determined by amplitudes higher/lower than ± 80 µV were excluded from further processing. The continuous EEG was subsequently epoched (from −125 to 1125 ms following stimulus onset), followed by averaging the epochs of each condition (minimum of 21 artifact-free epochs, range 21–60 epochs/ condition, for details see Supplementary Table S1) to derive the ERPs. Grand averages were calculated for each group, test time and condition. The interval chosen to analyse the visual N1 (128–216 ms) was determined by means of the interval between two subsequent global field power (GFP) sinks in the waveform, averaged over all conditions, test times and groups (Supplementary Figure S1). Mean amplitude values of the N1 interval were derived for electrode clusters over the left (LOT: PO7, O1’, P7, P5) and right (ROT: PO8, O2’, P8, P6) hemispheres.

Statistical analyses

SAS 9.4 (SAS Institute, Cary NC) procedure PROC MIX was used to compute linear mixed models (LMM) with the fixed factors hemisphere (LOT, ROT), test time (pre, post), condition (letters, false fonts, trained characters, untrained characters, novel font), group (ST, WT) and including the random intercept of each subject for the N1 mean amplitude. Separate LMMs were computed for training performance and task performance (percent hits and reaction time): these analyses included as fixed factors test time (post-training/ follow-up or pre-/post-training, respectively) and group (ST, WT), as well as the random intercept of each subject for the performance measure. Standardized residuals were calculated and values above 3 and below −3 were excluded from further analysis as outliers. To verify the assumptions of normality and homoscedasticity, visual inspection of QQ-plots and predicted versus residual plots were undertaken respectively. All analyses fulfilled the criteria. For significant interactions, post-hoc t tests were conducted. Uncorrected p values of post-hoc tests surviving multiple comparison correction using the Tukey-Kramer method are additionally marked with asterisks in the text (p_corr < .05*, p_corr < .01**, p_corr < .001***). All analyses were conducted with all 32 participants, including the three participants for whom the data of condition N were not available. To confirm the main findings also without these three participants, the LMM for the N1 amplitude was repeated for the reduced groups (ST: n = 13, WT: n = 16).

Results

Performance in the training

All subjects aside from one with an accuracy of 80% achieved more than 90% correct responses in the test level after the training, indicating successful learning of the characters. The ST and WT groups differed regarding the time spent for the training, whereby the ST group needed more time to successfully learn the character–speech sound associations compared to the WT group (p = .004). Moreover, the WT group achieved an overall higher accuracy directly after the training compared to the ST group (p = .013). However, this difference disappeared in the follow-up session where both groups showed equally high accuracies in the final training test level and the two separate character discrimination tests (see Table 1).

The LMM revealed a main effect of test time, F(1, 28) = 19.26, p < .0001, indicating that the performance in character knowledge was higher directly after the training than in the follow-up session after a few days over both groups. Furthermore, a group difference was detected, F(1, 28) = 7.59, p = .0102, indicating that the WT group achieved a better performance in the test level.

Performance in the one-back repetition detection task

The two groups did not differ regarding the number of hits, false alarms, omissions and reaction time in the one-back character detection task neither before nor after the computerized training (Table 1). The LMM revealed a main effect of time for the percentage of hits, F(1,30) = 6.67, p = .0149, but not for reaction time, thus indicating that the performance was higher before than after the training in both groups. The reduced accuracy after training may be explained by somewhat diminished attention to detect the character repetitions at the end of the rather long EEG session in both groups. No significant group effect or interaction of test time and group was found in any measure.

N1 amplitude analysis

The LMM revealed the main effects of hemisphere, F(1, 555) = 5.27, p = .0221, and condition, F(4, 555) = 19.061, p < .0001, indicating an overall more pronounced N1 amplitude in the left hemisphere and major differences between conditions. More importantly, a significant interaction of test time and condition, F(4, 555) = 3.10, p = .0153 pointed to condition-specific training effects (Figure 3 and Figure 4). Post-hoc tests revealed significant effects of test time only for the trained characters (p = .0004*). None of the other conditions showed a training effect (post vs. pre) difference at a trend level (all p > .25). Before the training, a significant condition difference was found mainly in the form of a reduced negativity for L and FF compared to the other conditions, but no difference was found between TC and UC characters: L vs. TC (p = .0116), L vs. UC (p = .0048), L vs. N (p = .0614), FF vs. TC (p = .0023), FF vs. UC (p = .0008*) and FF vs. N (p = .0169). The condition differences were more pronounced after training with the strongest negativity to TC after training in comparison to all other conditions, TC vs. UC (p = .0013*), TC vs. N (p = .0041), TC vs. FF (p = .0003**), TC vs. L (p < .0001***). L and FF exhibited least pronounced negativities similar to the pre-training data: L vs. UC (p = .0002**), L vs. N (p = .0001**), FF vs. UC (p = .0002**).

Figure 3.

Training effects on the N1 ERP (128–216 ms) shown in waveforms and topographical maps.

Figure 4.

Interaction plot of the N1 training effect.

Separate additional analyses of LOT and ROT (waveforms, Figure 3A) supported the above-reported effects with significant or marginally significant interactions of test time and condition, LOT: F(4, 258) = 2.22, p < .0678; ROT F(4, 259) = 4.46, p < .0017, and main effects of condition, LOT: F(4, 258) = 22.99, p < .0001; ROT F(4, 259) = 26.27, p < .0001. Of note, the additional LMM analysis with the reduced groups (ST: n = 13, WT: n = 16) confirmed the main effects of condition, F(4, 510) = 16.31, p < .0001, hemisphere, F(1, 510) = 7.78, p = .0055, and the interaction of condition and test time, F(4, 510) = 3.11, p = .0150, of the main analyses. No significant group effect or interaction of test time and group was found in any analysis (all F < 1.2, p > .25). Even though the data (supplementary Figure S2) suggested stronger N1 training effects for the ST group, the direct comparison with the WT group at the selected electrode clusters and the analyzed N1 interval yielded no significance.

Discussion

Training character–syllable and character–word associations

Here, we examine how a short training of novel characters modulates the visual N1 ERP regarding amplitude and lateralization while controlling for potential effects of time and repetition. The adult participants successfully learned a set of 12 character–syllable or character–word associations within less than 1.5 hours of training. Importantly, the character–syllable and character–word training induced an expertise effect as reflected in a bilateral increase of the visual N1 ERP. The behavioral training data showed that the word training group learned faster and performed better in the final test level after the training. This indicates that learning the associations between novel characters and concrete nouns was less challenging, probably because it is easier to build up mnemonic aids for concrete nouns than for abstract syllable sounds. However, these group differences disappeared at the time of the follow-up, when both groups achieved a similar performance in the test level as well as additional character discrimination and recognition tasks.

Training Effects on the Visual N1 tuning

The ERP analyses showed a pronounced training effect on the N1 amplitudes for trained characters. Only the trained characters revealed a change in amplitude with test time. While there was no amplitude difference between trained, untrained and novel characters before training, a pronounced amplitude difference was shown afterwards due to an increase in the negativity over the left and right occipito-temporal scalp for the trained condition only. In contrast to previous data showing an increased N1 after repeated presentation of symbol strings (Brem et al., 2005), neither the repeated (FF, UC) nor the novel (N) characters showed a change in amplitude between test times, thus indicating that repetition and time effects cannot explain the modulations seen for the trained characters; rather, learning novel associations between characters and phonological and/or lexical information along with gaining expertise caused the increase in the N1 amplitude. This result is strongly in line with previous studies showing increased visual N1 amplitudes for visual categories of specific expertise in adults (Bentin et al., 1999; Brem et al., 2005; Curran et al., 2002; Maurer et al., 2005; Rossion et al., 2003; Wong et al., 2005). The results also nicely match the effects of gaining expertise for print on the visual N1 when children learn to read (Brem et al., 2010, 2013; Cao et al., 2011; Maurer et al., 2006).

Interestingly, the N1 expertise effect did not differ between participants’ training character–word or character–syllable associations, neither in the strength of the training effect nor in the lateralization of the maximal amplitude over LOT and ROT. The absence of a lateralization effect seems to partly contradict the results reported in the articles by the group of Yoncheva, Maurer and colleagues. In their study, both an explicit reading task (Yoncheva et al., 2010, 2015) and an implicit visual one-back detection (Maurer et al., 2010) task were used to assess training effects. Moreover, they compared two different types of learning instructions and were able to show that selective attention to grapheme–phoneme mappings resulted in a left lateralized modulation of the N1 amplitude, while a whole-word focus during training drove the right lateralized N1 effect (Yoncheva et al., 2010, 2015) in the explicit reading task. However, the analyses of the implicit one-back task resulted in predominantly right hemispheric training effects for the same sample and independent of the attentional focus during training (Maurer et al., 2010). Differences in the task design and especially in terms of whether explicit reading or implicit automatic processing is required may thus account for the differences in the reported training effects: for explicit reading tasks only, the impact of the attentional focus during learning may modulate the lateralization of the N1 effect depending on whether holistic or segmental processing is prioritized (Yoncheva et al., 2010, 2015), while learning in general may induce an amplitude increase as seen in implicit tasks and independent of learning instruction. Visual inspection of the statistical, topographic maps of trained characters for pooled (Figure 3B) and separate groups (Supplementary Figure S2), all show a bilateral distribution for the training effect with only a slight extension to the right hemisphere. The absence of a clearly right lateralized training effect seen here may be explained by differences in the implicit task designs between Maurer et al.’s (2010) and our study, such as randomized vs blocked and/or short (100 ms) vs long (683 ms) stimulus presentations. Given that in Maurer et al.’s study only whole-word items were presented, it is conceivable that the implicit presentation favored holistic processing. However, further studies are necessary to clarify the N1 lateralization during emerging expertise in more detail. Given the widely-accepted assumption that the generators of the visual N1 are located in the extrastriate cortex (Allison et al., 1994; Maurer et al., 2005), the bilateral increase of the N1 in our study converges with the results of an fMRI study showing an increase in the activity of bilateral fusiform cortices after phonological and semantic training (Xue et al., 2006). The absence of a difference between syllable and word training also confirmed that not lexical access or semantic processing but rather aspects common to both trainings such as increasing visual familiarity and learning phonological associations shaped the visual expertise effect. The present findings are thus compatible with the hypothesis of a phonologically-guided tuning of the ventral occipito-temporal cortex to print by perysilvian regions during reading acquisition (Brem et al., 2010; Sandak et al., 2004; Schlaggar & McCandliss, 2007).

N1 tuning to single letters in adults

In general, the more complex visual stimuli (Japanese, Chinese and Indian scripts) elicited more pronounced N1 amplitudes at both test times compared to simpler characters such as those of the familiar font and the matched false font. Interestingly, we found least pronounced N1 amplitudes for the highly familiar letters and the unfamiliar false font condition. The reduced activation to familiar letters contrasts to the pronounced N1 letter sensitivity effect in a previous study comparing the processing of Chinese characters, pseudo-characters and Latin letters in a group of native English readers (Wong et al., 2005) and contrasts with previous findings showing print sensitivity to letter strings not only in children but also adults (Bentin et al., 1999; Brem et al., 2006; Maurer et al., 2005; Schendan et al., 1998). On the one hand, differences in the task design and analysis may account for some divergence in our findings. While most previous studies presented the conditions in blocks and some applied a pre-stimulus baseline ERP correction (Wong et al., 2005), we used a randomized presentation without pre-stimulus baseline correction. ERPs derived from block-wise presentations may be more affected by changes in the mental/attentional states between conditions compared to randomized presentation, while the application of a pre-stimulus baseline correction may have enhanced such state-dependent differences. On the other hand, our finding of reduced letter sensitivity in expert adult readers nicely corresponds to the hypothesis of an inverse U-shaped, expertise-dependent modulation of the strength of activity within the left ventral occipito-temporal cortex to print predicted by the interactive account model (Price & Devlin, 2011). Accordingly, the very high expertise expected for familiar letters in adult readers would result in a very low or even absent prediction error and correspondingly low activation given the precise, strong matching of top-down predictions and the visual feedforward input. Similarly, one would expect only low activation due to absent expertise/prediction errors for false fonts. This explanation is also supported by own work, showing that the N1 sensitivity to letter strings in adults is usually weaker compared to less experienced groups such as children learning to read (Brem et al., 2009; Maurer et al., 2006). Moreover, single letters may reach a higher level of expertise compared to printed words, resulting in an even lower prediction error. Finally, it is likely that the amplitude differences for L and/or FF vs. TC, UC and N before the training or vs. UC and N after the training reflect a difference in the physical appearance and visual complexity of the chosen foreign scripts (Japanese, Chinese, Indian) compared to the simpler Latin letters and matched FF.

Conclusion

To conclude, a short character–speech sound or character–word training within a single experimental session can induce expertise effects to a novel font in healthy normal reading adults and is reflected by an overall increase of the visual N1 over the occipito-temporal scalp. Together with previous training studies in adults, our data show that the changes in automatic script processing and the corresponding N1 amplitude with increasing visual expertise to novel characters is mainly driven by learning phonological associations.

Supplemental material

Supplemental Material, JBD727871_supplementary_table - Increasing expertise to a novel script modulates the visual N1 ERP in healthy adults

Supplemental Material, JBD727871_supplementary_table for Increasing expertise to a novel script modulates the visual N1 ERP in healthy adults by Urs Maurer, Catherine McBride, Silvia Brem, Eliane Hunkeler, Markus Mächler, Jens Kronschnabel, Iliana Irini Karipidis, Georgette Pleisch, and Daniel Brandeis in International Journal of Behavioral Development

Supplemental material

Supplemental Material, JBD727871_supplementary_figures - Increasing expertise to a novel script modulates the visual N1 ERP in healthy adults

Supplemental Material, JBD727871_supplementary_figures for Increasing expertise to a novel script modulates the visual N1 ERP in healthy adults by Urs Maurer, Catherine McBride, Silvia Brem, Eliane Hunkeler, Markus Mächler, Jens Kronschnabel, Iliana Irini Karipidis, Georgette Pleisch, and Daniel Brandeis in International Journal of Behavioral Development

Footnotes

Acknowledgment

We thank Maja Schneebeli and Antonia Bak for their help with data analyses and recordings and Alexander Roth for statistical counseling. Furthermore, we thank Ulla Richardson and her team of the University of Jyväskylä for support with the training software.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is supported by Swiss National Science Foundation (grant #32003B_141201) and the Hartman Müller-Foundation (grant: 1912).

Supplemental material

Supplementary material for this article is available online.

References

Allison

McCarthy

Nobre

Puce

Belger

(1994). Human extrastriate visual cortex and the perception of faces, words, numbers, and colors. Cerebral Cortex, 4, 544–554.

Araújo

Bramão

Faísca

Petersson

K. M.

Reis

(2012). Electrophysiological correlates of impaired reading in dyslexic pre-adolescent children. Brain and Cognition, 79, 79–88.

Auer

Gruber

Mayringer

Wimmer

. (2005). Salzburger Lese-Screening für die Klassenstufen 5–8 (SLS 5-8). Bern: Hans Huber.

Bentin

Mouchetant-Rostaing

Giard

M. H.

Echallier

J. F.

Pernier

(1999). ERP manifestations of processing printed words at different psycholinguistic levels: Time course and scalp distribution. Journal of Cognitive Neuroscience, 11, 235–260.

Brem

Bach

Kucian

Kujala

J. V.

Guttorm

T. K.

Martin

… Richardson

. (2010). Brain sensitivity to print emerges when children learn letter-speech sound correspondences. Proceedings of the National Academy of Sciences U S A, 107, 7939–7944. doi:10.1073/pnas.0904402107

Brem

Bach

Kujala

J. V.

Maurer

Lyytinen

Richardson

Brandeis

(2013). An electrophysiological study of print processing in kindergarten: The contribution of the visual n1 as a predictor of reading outcome. Developmental Neuropsychology, 38, 567–594. doi:10.1080/87565641.2013.828729

Brem

Bucher

Halder

Summers

Dietrich

Martin

Brandeis

(2006). Evidence for developmental changes in the visual word processing network beyond adolescence. Neuroimage, 29, 822–837.

Brem

Halder

Bucher

Summers

Martin

Brandeis

. (2009). Tuning of the visual word processing system: distinct developmental ERP and fMRI Þffects. Human Brain Mapping, 30, 1833–1844.

Brem

Lang-Dullenkopf

Maurer

Halder

Bucher

Brandeis

(2005). Neurophysiological signs of rapidly emerging visual expertise for symbol strings. Neuroreport, 16, 45–48.

10.

Busey

T. A.

Vanderkolk

J. R.

(2005). Behavioral and electrophysiological evidence for configural processing in fingerprint experts. [Research Support, U.S. Gov’t, P.H.S.]. Vision Research, 45, 431–448. doi:10.1016/j.visres.2004.08.021

11.

Cao

Zhao

Lin

Weng

(2011). Left-lateralized early neurophysiological response for Chinese characters in young primary school children. Neuroscience Letters, 492, 165–169. doi:10.1016/j.neulet.2011.02.002

12.

Cohen

Dehaene

Naccache

Lehericy

Dehaene-Lambertz

Henaff

M. A.

Michel

(2000). The visual word form area: Spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients. Brain, 123, 291–307.

13.

Curran

Tanaka

J. W.

Weiskopf

D. M.

(2002). An electrophysiological comparison of visual categorization and recognition memory. Cognitive, Affective, & Behavioral Neuroscience, 2, 1–18.

14.

Eberhard-Moscicka

A. K.

Jost

L. B.

Raith

Maurer

(2015). Neurocognitive mechanisms of learning to read: Print tuning in beginning readers related to word-reading fluency and semantics but not phonology. Developmental Science, 18, 106–118. doi:10.1111/desc.12189

15.

Fraga González

Žarić

Tijms

Bonte

Blomert

van der Molen

M. W.

(2014). Brain-potential analysis of visual word recognition in dyslexics and typically reading children. Frontiers in Human Neuroscience, 8. doi: 10.3389/fnhum.2014.00474

16.

Gauthier

Curran

Curby

K. M.

Collins

(2003). Perceptual interference supports a non-modular account of face processing. Nature Neuroscience, 6, 428–432.

17.

Hashimoto

Sakai

K. L.

(2004). Learning letters in adulthood: Direct visualization of cortical plasticity for forming a new link between orthography and phonology. Neuron, 42, 311–322. doi:10.1016/S0896-6273(04)00196-5

18.

James

K. H.

(2010). Sensori-motor experience leads to changes in visual processing in the developing brain. Developmental Science, 13, 279–288. doi:10.1111/j.1467-7687.2009.00883.x

19.

Karipidis

I. I.

Pleisch

Röthlisberger

Hofstetter

Dornbierer

Stämpfli

Brem

(2017). Neural initialization of audiovisual integration in prereaders at varying risk for developmental dyslexia. Human Brain Mapping, 38, 1038–1055. doi:10.1002/hbm.23437

20.

Kim

K. H.

Kim

J. H.

(2006). Comparison of spatiotemporal cortical activation pattern during visual perception of Korean, English, Chinese words: An event-related potential study. Neuroscience Letters, 394, 227–232.

21.

Lyytinen

Erskine

Kujala

Ojanen

Richardson

(2009). In search of a science-based application: A learning tool for reading acquisition. Scandinavian Journal of Psychology, 50, 668–675. doi:10.1111/j.1467-9450.2009.00791.x

22.

Lyytinen

Ronimus

Alanko

Poikkeus

A. M.

Taanila

(2007). Early identification of dyslexia and the use of computer game-based practice to support reading acquisition. Nordic Psychology, 59, 109–126.

23.

Maurer

Blau

V. C.

Yoncheva

Y. N.

McCandliss

B. D.

(2010). Development of visual expertise for reading: Rapid emergence of visual familiarity for an artificial script. Developmental Neuropsychology, 35, 404–422. doi:10.1080/87565641.2010.480916

24.

Maurer

Brem

Bucher

Brandeis

(2005). Emerging neurophysiological specialization for letter strings. Journal of Cognitive Neuroscience, 17, 1532–1552.

25.

Maurer

Brem

Kranz

Bucher

Benz

Halder

… Brandeis

(2006). Coarse neural tuning for print peaks when children learn to read. Neuroimage, 33, 749–758.

26.

Maurer

Zevin

J. D.

McCandliss

B. D.

(2008). Left-lateralized N170 effects of visual expertise in reading: Evidence from Japanese syllabic and logographic scripts. Journal of Cognitive Neuroscience, 20, 1878–1891. doi:10.1162/jocn.2008.20125

27.

McCandliss

B. D.

Posner

M. I.

Givon

(1997). Brain plasticity in learning visual words. Cognitive Psychology, 33, 88–110. doi:10.1006/cogp.1997.0661

28.

Moll

Landerl

(2010). Lese- und Rechtschreibtest (SLRT-II) Weiterentwicklung des Salzburger Lese- und Rechtschreibtests (SLRT). Bern, Switzerland: Verlag Hans Huber.

29.

Nobre

A. C.

Allison

McCarthy

(1994). Word recognition in the human inferior temporal lobe. Nature, 372, 260–263.

30.

Park

Chiang

Brannon

E. M.

Woldorff

M. G

. (2014). Experience-dependent hemispheric specialization of letters and numbers is revealed in early visual processing. Journal of Cognitive Neuroscience, 26, 2239–2249. doi:10.1162/jocn_a_00621

31.

Pegado

Comerlato

Ventura

Jobert

Nakamura

Buiatti

… Dehaene

. (2014). Timing the impact of literacy on visual processing. Proceedings of the National Academy of Sciences U S A, 111, E5233–E5242. doi:10.1073/pnas.1417347111

32.

Price

C. J.

Devlin

J. T.

(2011). The Interactive Account of ventral occipitotemporal contributions to reading. Trends in Cognitive Sciences, 15, 246–253.

33.

Proverbio

A. M.

Vecchi

Zani

(2004). From orthography to phonetics: ERP measures of grapheme-to-phoneme conversion mechanisms in reading. Journal of Cognitive Neuroscience, 16, 301–317.

34.

Rossion

Gauthier

Goffaux

Tarr

M. J.

Crommelinck

(2002). Expertise training with novel objects leads to left-lateralized facelike electrophysiological responses. Psychological Science, 13, 250–257.

35.

Rossion

Joyce

C. A.

Cottrell

G. W.

Tarr

M. J.

(2003). Early lateralization and orientation tuning for face, word, and object processing in the visual cortex. Neuroimage, 20, 1609–1624.

36.

Saine

N. L.

Lerkkanen

M. K.

Ahonen

Tolvanen

Lyytinen

(2011). Computer-assisted remedial reading intervention for school beginners at risk for reading disability. Child Development, 82, 1013–1028. doi:10.1111/j.1467-8624.2011.01580.x

37.

Sandak

Mencl

W. E.

Frost

S. J.

Rueckl

J. G.

Katz

Moore

D. L.

… Pugh

K. R

. (2004). The neurobiology of adaptive learning in reading: A contrast of different training conditions. Cognitive, Affective, & Behavioral Neuroscience, 4, 67–88.

38.

Schendan

H. E.

Ganis

Kutas

(1998). Neurophysiological evidence for visual perceptual categorization of words and faces within 150 ms. Psychophysiology, 35, 240–251.

39.

Schlaggar

B. L.

McCandliss

B. D.

(2007). Development of neural systems for reading. Annual Review of Neuroscience, 30, 475–503. doi:10.1146/annurev.neuro.28.061604.135645

40.

Shirahama

Ohta

Takashima

Matsushima

Okubo

(2004). Magnetic brain activity elicited by visually presented symbols and Japanese characters. Neuroreport, 15, 771–775.

41.

Tanaka

J. W.

Curran

(2001). A neural basis for expert object recognition. Psychological Science, 12, 43–47.

42.

Vinckier

Dehaene

Jobert

Dubus

J. P.

Sigman

Cohen

(2007). Hierarchical coding of letter strings in the ventral stream: Dissecting the inner organization of the visual word-form system. Neuron, 55, 143–156.

43.

Wong

A. C.

Gauthier

Woroch

DeBuse

Curran

(2005). An early electrophysiological response associated with expertise in letter perception. Cognitive, Affective, & Behavioral Neuroscience, 5, 306–318.

44.

Xue

Chen

Jin

Dong

(2006). Language experience shapes fusiform activation when processing a logographic artificial language: An fMRI training study. Neuroimage, 31, 1315–1326.

45.

Yoncheva

Y. N.

Blau

V. C.

Maurer

McCandliss

B. D.

(2010). Attentional focus during learning impacts N170 ERP responses to an artificial script. Developmental Neuropsychology, 35, 423–445. doi:10.1080/87565641.2010.480918

46.

Yoncheva

Y. N.

Somandepalli

Reiss

P. T.

Kelly

Di Martino

Lazar

… Castellanos

F. X

. (2016). Mode of anisotropy reveals global diffusion alterations in attention-deficit/hyperactivity disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 55, 137–145. doi:10.1016/j.jaac.2015.11.011

47.

Yoncheva

Y. N.

Wise

McCandliss

(2015). Hemispheric specialization for visual words is shaped by attention to sublexical units during initial learning. Brain and Language, 145–146, 23–33. doi:10.1016/j.bandl.2015.04.001

48.

Zhao

Kipp

Gaspar

Maurer

Weng

Mecklinger

. (2014). Fine neural tuning for orthographic properties of words emerges early in children reading alphabetic script. Journal of Cognitive Neuroscience, 26, 2431–2442. doi:10.1162/jocn_a_00660

49.

Ziegler

J. C.

Goswami

(2005). Reading acquisition, developmental dyslexia, and skilled reading across languages: A psycholinguistic grain size theory. Psychological Bulletin, 131, 3–29.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.42 MB

0.17 MB