Brain & Language 145–146 (2015) 42–52

Contents lists available at ScienceDirect

Brain & Language journal homepage: www.elsevier.com/locate/b&l

Individual differences in involvement of the visual object recognition system during visual word recognition Sarah Laszlo a,b,⇑, Elizabeth Sacchi a a b

Department of Psychology, Binghamton University, United States Program in Linguistics, Binghamton University, United States

a r t i c l e

i n f o

Article history: Received 8 September 2014 Accepted 25 March 2015 Available online 16 May 2015 Keywords: Event-related potentials Visual word recognition Visual object recognition Individual differences

a b s t r a c t Individuals with dyslexia often evince reduced activation during reading in left hemisphere (LH) language regions. This can be observed along with increased activation in the right hemisphere (RH), especially in areas associated with object recognition – a pattern referred to as RH compensation. The mechanisms of RH compensation are relatively unclear. We hypothesize that RH compensation occurs when the RH object recognition system is called upon to supplement an underperforming LH visual word form recognition system. We tested this by collecting ERPs while participants with a range of reading abilities viewed words, objects, and word/object ambiguous items (e.g., ‘‘SMILE’’ shaped like a smile). Less experienced readers differentiate words, objects, and ambiguous items less strongly, especially over the RH. We suggest that this lack of differentiation may have negative consequences for dyslexic individuals demonstrating RH compensation. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction Specific reading impairment (also called dyslexia) is, by far, the most common learning disorder, with estimates of incidence reaching 12% of the population (e.g., Lindgren, De Renzi, & Richman, 1985). Given the fundamental necessity of literacy in our society, low levels of reading achievement can lead to a variety of difficulties in daily life that persist across the lifespan (e.g., Boetsch, Green, & Pennington, 1996), in addition to the obvious scholastic disadvantages. Consequently, there is a large literature addressed to understanding the cognitive and neural problems that are associated with dyslexia. In the cognitive domain, focus is on understanding what particular sub-skills of reading seem to be especially critical for the acquisition of expert reading skills. Phonological awareness (i.e., a metalinguistic awareness of the sound structure of the language referring to the ability to identify and manipulate the phonological units of words – for example, phonemes, syllables, and rhymes) is frequently highlighted as an especially strong candidate (see extensive review in National Reading Panel, 2000). Extensive behavioral work has further demonstrated that reading is a complex process that can be impaired due to problems with a number of subskills in addition

⇑ Corresponding author at: Department of Psychology, 4400 Vestal Parkway East, Binghamton, NY 13902, United States. E-mail address: [email protected] (S. Laszlo). http://dx.doi.org/10.1016/j.bandl.2015.03.009 0093-934X/Ó 2015 Elsevier Inc. All rights reserved.

to phonological awareness, such as visual attention (e.g., Bosse, Tainturier, & Valdois, 2007), orthographic analysis (e.g., Stanovich & West, 1989; Stanovich, West, & Cunningham, 1991), vocabulary knowledge (review in NRP, 2000), and fluent recognition of text (e.g., Rashotte & Torgesen, 1985), In the neural domain, a large number of studies have focused on understanding how the brains of individuals with dyslexia differ from those of individuals who read normally. Results from this work suggest several common characteristics of the functional brain organization of dyslexics. First, dyslexics seem to evince reduced functional connectivity in the left cerebral hemisphere (e.g., Keller & Just, 2009; Steinbrink et al., 2008). This reduced connectivity seems to be related to, second, reduced left hemisphere activation during phonological processing tasks, especially in the left inferior frontal gyrus and temporo-parietal areas (e.g., Hoeft et al., 2007; Keller & Just, 2009; Shaywitz & Shaywitz, 2005; Temple et al., 2003). Reduced activity in the LH is also observed in the occipitotemporal system during word reading or reading related tasks (e.g., Paulesu et al., 2001) – though not all changes in LH activation in dyslexia are reductions; activation in the left inferior frontal gyrus has been shown to increase in some dyslexic individuals (Pugh et al., 2010; Shaywitz et al., 1998) Finally, seemingly in compensation for reduced LH effectiveness, poor readers – even those that are not dyslexic – have been shown to exhibit more activation in right hemisphere occipitotemporal (OT) regions (Shaywitz & Shaywitz, 2005) than better readers. Right hemisphere

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

compensation is in its extreme in alexia, where seemingly compensatory activity in the right OT fusiform gyrus has been observed when the left OT fusiform region is actually damaged (e.g., Cohen et al., 2003). Here, we are interested in two, related, questions. First, why is RH OT cortex specifically, of all the brain, a region that is frequently observed to contribute to RH compensation in dyslexia? Second why is this RH OT compensation seemingly maladaptive? That is, why do children who use the RH OT more during reading demonstrate lower reading ability than children who use it less? Why is its use not helping to address connectivity/activation problems in the LH analog? That is, it is certainly imaginable that, in the face of an abnormally developing LH, use of the RH could be beneficial, with individuals who use it more demonstrating stronger reading ability than those that use it less. However, this is not what is observed (Shaywitz & Shaywitz, 2005), and beginning to understand why additional use of the RH does not seem to protect reading skill in the face of problems in the LH is one goal of this paper. 1.1. Why the RH OT cortex? When RH OT compensation occurs, why does it occur in RH OT cortex specifically, and not in a RH analogue of any of the other LH regions involved in the reading network (a number of which are involved in normal reading already)? When the LH network is actually damaged, as in alexia, for example, compensation can instead take place in the RH fusiform gyrus (Cohen et al., 2003). It is our hypothesis that the typical function of the RH OT cortex is related enough to the typical function of the LH OT cortex to be recruited to assist a sparsely connected LH – to a greater degree than other RH analogues of LH language regions. We will argue that the reason that RH OT cortex, in particular, is recruited in RH compensation is that this compensation reflects recruitment of the visual object recognition system for dealing with visual wordforms. There is a fairly strong literature suggesting that the LH OT cortex is involved in the decoding of visual word forms into more abstract orthographic features (e.g., Cohen & Dehaene, 2004; Dehaene & Cohen, 2011). Though there is spirited debate pertaining to how specifically the LH OT is involved in orthographic analysis (Price & Devlin, 2003, 2011), especially regarding how and whether this region might perform other functions, the evidence is strong that it does at least contribute to orthographic analysis. The result of orthographic analysis is the extraction of abstract orthographic features from visual percepts; these representations can then be processed at higher linguistic levels such as semantics or lexicality (see Grainger & Holcomb, 2009; Laszlo & Federmeier, 2014; Laszlo & Plaut, 2012). The RH OT seems to undertake a related function with related results, but in a different domain. Specifically, we note that this region seems to be involved in visual object recognition (e.g., Grill-Spector, Kourtzi, & Kanwisher, 2001; Haxby et al., 1991; Spiridon & Kanwisher, 2002). Visual object recognition involves the extraction of more abstract representations from visual percepts – representations that may include or eventually be linked with category labels or other verbal information (see review in Tarr & Vuong, 2002). Thus, while in the LH OT visual percepts are converted to abstract features for further processing, in the RH OT a similar process seems to take place, only based on visual objects instead of visual wordforms. Clearly, the functions of the two regions are related. In fact, it is likely that the specialization of the LH OT for orthographic analysis and the specialization of the RH OT for object recognition are only relative specializations, with both areas likely being involved to some extent with both processes (e.g., Price & Devlin, 2003; Seghier & Price, 2011) This is consistent with findings in the literature that demonstrate that

43

the RH OT system is called upon to support the LH OT system for text processing in various situations, such as when text is presented vertically instead of horizontally (Cai, Paulignan, Brysbaert, Ibarrola, & Nazir, 2010) or in children who have not yet learned how to read (Maurer, Brem, Bucher, & Brandeis, 2005). Indeed, it is thought that the RH’s visuospatial expertise is partially responsible for its heightened involvement in processing Chinese text relative to English and other alphabetic scripts (e.g., Liu, Dunlap, Fiez, & Perfetti, 2007). It is thought that this heightened RH involvement may be due to the ‘‘global’’ (low spatial frequency) nature of the processing that takes place in logographic languages as opposed to ‘‘local’’ (high spatial frequency) information more critical for decoding of alphabetic languages (e.g., Mei et al., 2014). Here, we will directly test the possibility that RH OT involvement in text processing represents use of the object recognition system to support the LH word recognition system, and whether degree of use of the RH OT in this role varies according to individual reading ability. We will do this by recording ERPs while participants who vary in their vocabulary, verbal fluency, and exposure to print view words, objects, and word/object ambiguous items (e.g., the word SMILE shaped like a smile, see Fig. 1 for examples and Appendix 1 for the full set). This experimental design enables the exploration of several questions. First, do individual weaker readers display less specialization for print than stronger readers, as evidenced by less differentiated responses to the words and objects? Second, how do individuals respond to the novel, ambiguous items, and how does reading ability impact the processing of those items? Third, how do any observed reading ability effects interact with hemisphere of processing? We will explore each of these questions in our analysis. 1.2. Characteristics of RH use in reading Individuals who evince more RH OT activation also tend to be weaker readers (Shaywitz & Shaywitz, 2005), and consequently, strengthening LH reliance is a stated target of multiple reading interventions (Richards & Berninger, 2008; Shaywitz et al., 2004). But more broadly, involvement of the RH in language comprehension is not only not maladaptive, it is normal and essential. Thus, as we consider hemispheric asymmetries in print specialization and how those interact with reading ability, it is important to note that RH involvement in reading is not always a sign of disorder – in fact the RH is clearly involved in many advanced reading functions in normal readers, such as integrating incoming text with sentence or discourse context (e.g., Federmeier, 2007; Federmeier & Kutas, 1999; Wlotko & Federmeier, 2007), resolving lexical and semantic ambiguities (Kandhadai & Federmeier, 2008, 2010), and interpreting non-literal language, as in jokes (Coulson & Williams, 2005). Further, in reading disorders besides dyslexia, such as alexia, RH compensation is associated with recovery (e.g., Cohen et al., 2003). Indeed, the apparent over-activation of the RH OT in dyslexia may only be an artifact of comparing its activity with an underactivating LH analog (see Shaywitz & Shaywitz, 2008) – that is, it may not be that the RH OT is actually working more at an absolute level in dyslexic individuals than in normally developing readers; rather it may just be working proportionally more. Similarly, the finding that individuals who rely more on the RH OT system during reading also are poorer readers (Shaywitz & Shaywitz, 2005) could potentially be explained with the suggestion that increased RH reliance is always or mostly associated with greater problems in the LH, meaning that RH compensation is not the cause of poorer reading ability but only a symptom of the underlying problems in the LH. Nevertheless, it is still the case that many established reading remediations aim to increase reliance on LH areas during reading, precisely because overuse of the RH is seen as maladaptive

44

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

Fig. 1. Sample stimuli. Sample words (left), objects (center), and word/object ambiguous items (right). All items were presented on the same 480  480 pixel gray background.

(Richards & Berninger, 2008; Shaywitz et al., 2004) – even though alternate explanations exist. Why might the perception that over-reliance on the RH is maladaptive be justified? We propose that even a relative shift to more RH OT processing of text could be maladaptive in the absence of a properly functioning LH analog if the RH OT cannot produce orthographic representations of text that are as detailed and sharp as those of the LH. When the LH system is working properly, involvement of the RH OT is not problematic, as the LH OT system can share its information and ‘‘clean up’’ (e.g., Joanisse & Seidenberg, 1999 – here, and in related computational work, ‘‘clean up’’ refers to a sub-population of computational units or neurons working to sharpen a noisy representation within its layer of representation) any noisy or sparse orthographic representations computed by the RH OT. However, when the LH system is not able to perform this maintenance, and RH OT orthographic representations dominate (or cannot be inhibited), the entire remainder of the comprehension system that relies on those representations may be disadvantaged. What reason is there to think that RH OT representations of text would be less viable than those of the LH? One crucial reason is that the RH OT is known to be less sensitive to high spatial frequencies than the LH analog (e.g., Kitterle, Christman, & Hellige, 1990; see review in Flevaris, Martinez, & Hillyard, 2014). The differences between letters in printed text are precisely the kind of high spatial frequency information that the RH does not excel at extracting. Indeed, one suggested means to make reading easier for dyslexic individuals is simply to space letters more widely apart (review in McCandliss, 2012); a procedure which reduces the spatial frequency of print. Again, the RH OT’s relative insensitivity to high spatial frequencies may not be especially or devastatingly problematic when the LH OT is functioning at full strength, and can sharpen representations computed by the RH. But when the LH OT is not functioning properly, noisy orthographic representations may disadvantage the entire rest of the system. In readers whose LH develops normally, the noisy RH orthographic representations may become inhibited (or weakly weighted) through learning. However, in readers whose LH develops with sparse connectivity, LH orthographic representations may not be of a

quality that is sufficiently high to result in inhibition or clean up of orthographic information provided by the RH. Lack of RH sensitivity to spatial frequency is not the only reason to believe that print processing may be more effective when it occurs in the LH. The LH OT cortex is known to have connections with other LH regions associated with phonemic and semantic processing, such as the left inferior frontal cortex (Van der Mark et al., 2011). These connections to the rest of the language comprehension system confer a privilege on the LH OT system that the RH analog does not share. Indeed, under some theoretical frameworks, such as the phonological mapping hypothesis, it is proposed that the reason that the LH is seemingly dominant for many parts of print processing is precisely of its privileged connections with LH regions devoted to auditory processing of language (see Maurer & McCandliss, 2008). To the extent that RH OT involvement in text processing is in fact problematic, and interventions targeting its reduction are well-aimed, it is important to understand better the function it provides when it is recruited for reading, and the nature of the representations it computes. This is what we will set out to accomplish here, especially in our examination of RH responses to word/object ambiguous items and any possible individual differences therein. 1.3. Ambiguous items and inter-hemispheric communication The word/object ambiguous items we consider here have not, to our knowledge, been studied before in the neuroimaging literature broadly, or in the ERP literature more specifically. However, these items present an interesting challenge to the system, as they have the physical properties of both text and objects. Should they be routed primarily to the RH OT system, for processing as objects, or primarily to the LH OT system, for orthographic analysis? And does this routing depend on reading ability? In response to the ambiguous items, it seems likely that two parallel processing streams are initiated. In the RH, the lower spatial frequency, object-like contours of the ambiguous items are processed simultaneously with, in the LH, the extraction of abstract orthography from the higher spatial frequency orthographic patterns also

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

present in the ambiguous items. In adult readers whose LH systems are more densely and appropriately connected, however, any contribution from the RH OT to the ongoing and simultaneous computation of abstract orthography occurring in the LH may be relatively inhibited (or, equivalently, weakly weighted), as a result of relatively poor information about orthographic features being provided by that system throughout development (see an example of how this type dynamic could develop in Harm & Seidenberg, 2004). In contrast, the fidelity of orthographic information may not be as different across hemispheres in individuals whose LH systems have developed with sparse connections, meaning that RH orthographic information would not have been learned to be as strongly inhibited (or weakly weighted) in the poorer readers. Intrahemispheric differences in efficiency of processing the relatively high spatial frequencies in orthography over the course of development, then, may provide a context wherein better adult readers are more successful at interhemispheric inhibition of information about the ambiguous objects coming in from the RH. This is what we mean to suggest constitutes the differential ‘‘routing’’ of the ambiguous items to the LH in the better readers. 1.4. The N170 Our measures of interest will be the amplitude and distribution of the N170 ERP component. The N170 (also known as the visual N1) is a negativity with a narrow scalp distribution that is primarily maximal over a restricted range of lateral occipital sites, peaking around 170 ms post stimulus onset (review in Rossion & Jacques, 2012). The function of the N170 seems to be the ‘‘organization of visual cues into structured patterns’’ (Rossion & Caharel, 2011, p.1308) – essentially the same function we have ascribed to the LH and RH OT. This is not surprising, as the OT cortices have been identified as generators of the N170 (Rossion & Jacques, 2012). And, indeed, the lateralization of the N170 seems to vary depending on what type of stimulus is being presented, with faces eliciting strongly right lateralized N170s (e.g., Rossion, Joyce, Cottrell, & Tarr, 2003), wordforms eliciting left lateralized N170s (e.g., Rossion et al., 2003), and other objects eliciting relatively bilateral N170s (Rossion et al., 2003). The N170 is also characterized by its sensitivity to the level of expertise an individual has with a stimulus, with more familiar items generating more positive potentials than less familiar items (Tanaka & Curran, 2001). Here, note that all visual stimuli elicit some N170 activity, it is only the relative lateralization that varies across stimulus types. Insofar as N170s elicited by print are relatively left lateralized, it has been hypothesized that this LH dominance is a result of the LH OT cortex’s relatively privileged connections with other LH areas dominant for the auditory processing of language (the phonological mapping hypothesis, Maurer & McCandliss, 2008). Dyslexic individuals display relatively less N170 left lateralization in response to words than do non-dyslexics (Mahé, Bonnefond, Gavens, Dufour, & Doignon-Camus, 2012), consistent with the finding of RH OT compensation in dyslexia. Other individual factors contribute to N170 lateralization as well. Children who have not yet learned to read do not evince strongly lateralized N170s in response to print (Maurer et al., 2005), suggesting that left lateralization of the N170 to words is a phenomenon that develops with literacy – and in agreement with our speculation regarding developmentally primed ‘‘routing’’ of the ambiguous items to the RH and LH OT systems as described above. Further, individuals who learn a novel form of type demonstrate more left-lateralized N170s when they are taught using a sub-lexical method than when they are taught using a whole-word method (Yoncheva, Blau, Maurer, & McCandliss, 2010). Finally, individuals who are expert at recognizing particular types of visual stimuli (e.g., faces, cars, birds, or dogs) may also demonstrate especially

45

large RH N170s to the categories of their expertise (e.g., Gauthier, Curran, Curby, & Collins, 2003; Rossion, Collins, Goffaux, & Curran, 2007; Tanaka & Curran, 2001). Taken as a whole, these findings are supportive of the idea that the functions of the LH and RH OT systems are flexible, and that the RH system might be called upon to support a sparsely connected LH system for the purpose of text processing. We will examine whether the N170s elicited to words and objects are more similar in weaker readers, especially over right hemisphere sites. Such a lack of specialization would suggest that, indeed, the RH object recognition system becomes relatively more involved in word recognition in these readers. More novelly, we will examine the N170 responses to word/object ambiguous items, to see whether weaker readers elicit N170s to ambiguous items that are more like those elicited by objects than in stronger readers. 2. Methods 2.1. Participants A total of 48 undergraduate students from the State University of New York at Binghamton and Broome County Community College participated in this study (29 female; mean age 19.86, age range 18–29); the latter were included to try to achieve greater variability in reading ability than would be expected within a University subject pool. Nevertheless, none of the participants had a formal diagnosis of dyslexia. Three of the original 48 participants were excluded due to excessive noise or ocular artifacts, leaving 45 participants to be included in analysis. All participants were compensated with either course credit or money. The Internal Review Board of Binghamton University approved the experimental procedures. All participants identified themselves as monolingual English speakers with little to no early life exposure to any second language. Participants also reported normal or corrected to normal vision as well as no history of psychiatric or neurological disorders. In addition, no participants reported taking any medication, prescription, non-prescription, or recreational, which would affect their cognitive processes. All participants were right handed, as indicated by a handedness questionnaire modeled after Oldfield (1971). 2.2. Reading assessments Each participant’s verbal fluency, vocabulary, and exposure to print were assessed before beginning the EEG portion of the experiment. Verbal fluency was assessed with a semantic membership task: participants were asked to list as many examples of each of 3 semantic categories (fruits, animals, and colors) as they could in 60 s. Verbal fluency is known to be significantly impaired in dyslexia (Reiter, Tucha, & Lange, 2005) and can even be used to clinically differentiate sub-forms of dyslexia (Cohen, Morgan, Vaughn, Riccio, & Hall, 1999) Vocabulary was assessed with a task requiring participants to match each of 30 words in a list with a printed definition. The 30 items included 10 each ‘‘easy’’, ‘‘medium’’, and ‘‘hard’’ GRE words. Vocabulary is central to reading comprehension, which some view as ‘the essence of reading’ (NRP, 2000, p. 4–1). Participants’ exposure to print was assessed with a magazine recognition task and an author recognition task; these materials were those developed beginning with Stanovich & West, 1989. In each recognition task, participants were asked to indicate which of a list of authors/magazines were real and which were not real. Exposure to print as measured by this instrument is known to be reliably related to orthographic analysis (e.g., Stanovich & West, 1989), which is further known to explain variability in reading

46

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

ability above and beyond that explained by phonological awareness (see Stanovich et al., 1991, see also Cunningham & Stanovich, 1991; Stanovich & Cunningham, 1992; West & Stanovich, 1991 for more information on the magazine and author recognition tasks); further, as mentioned in the introduction, LH OT dominance for print processing seems to only accrue as a function of experience with text (Maurer et al., 2005). That the magazine and author recognition tests are known to relate strongly and specifically to orthographic knowledge and experience (e.g., West & Stanovich, 1991) means that they are especially relevant here. 2.3. Materials Stimuli for the EEG portion of the experiment consisted of 6 categories of images, with 50 images per category for a total of 300 images. The 6 categories were words, outlined words, word/object ambiguous items, objects, faces, and animals. The word, object, and word/object ambiguous items were the critical categories; the other images served as fillers (faces, outlined words) or targets in the behavioral Go/Nogo task, which was to respond when an image of an animal was presented and not respond otherwise. All images were presented on the same uniform gray background sized 480  480 pixels. The word category consisted of 50 images of words presented in a black san serif font, 72 pt. The outlined words category consisted of 50 images of words in the same san serif font, in white with a black outline. The word/object ambiguous items were images of words in black shaped like the concept they represent (e.g., the word SMILE drawn to be shaped like a smile; see Fig. 1 for examples, see Appendix 1 for the full list). The ambiguous items were designed so that it would not be clear, from their physical form, whether to process them as visual word forms or as objects (i.e., the form of these images was meant to be intermediate between words and objects). Evidence that this was the case is depicted in Fig. 2, where it is clear that, on average, over a majority of the central scalp, the ambiguous items were treated intermediately between unambiguous words and unambiguous objects, eliciting intermediate waveforms. The remaining three categories (images of objects, faces, and animals) were all presented in black and white on the same gray background as the word-containing images. The three categories containing images of words (words, outlined words, and word/object ambiguous items) were matched for length, Coltheart’s N, orthographic frequency, bigram frequency, syllable frequency, and number of syllables using MCWord: An Orthographic Wordform Database (Medler & Binder, 2005) and SUBTLEXus (Brysbaert & New, 2009). Table 1 summarizes the means for each word type (words, outlined words, and ambiguous items) across each of these five lexical properties; the complete list of items is available in Appendix 1. There were no reliable differences between categories on any of the dimensions listed above. 2.4. Procedure Participants were seated in an electrically shielded, sound attenuated booth and positioned 75 cm away from a 24 inch LCD computer monitor with a resolution of 1920  1080. Before the EEG portion of the experiment began, participants were shown their own EEG record in real time and were given a presentation showing what muscle movements and blinks looked like on the EEG record. This presentation was given in order to inform participants why it was important to remain still and blink only at the times specified during the experiment. Following this presentation, participants were guided through a demonstration of the experiment, familiarizing them with the structure of the experiment, including times that were and were not acceptable for blinking

and eye movements. Participants were informed that their task was to press a green button on a gamepad as fast as they could with their right hand when they saw a picture of a non-human animal (a Go/NoGo task). Participants were told that they did not have to do anything when they saw pictures that were not animals. This behavioral Go/NoGo task was included to ensure that participants were attending to the stimuli being presented, but to also keep ERPs elicited in response to critical items (words, objects, and word–object ambiguous items) free of response potentials. Following the demonstration, the 300 images were presented in a random order for each participant. Each image was only presented once. The 300 trials were broken down into three blocks of 100 trials, and participants were given a break between each block. The total duration of the experiment was approximately 2 h. Trial structure was as follows. A 480  480 pixel gray box was continuously present on the screen over a black background. At the beginning of each trial, a black fixation cross was presented in the center of the gray box for a random duration jittered around a mean of 500 ms. After this fixation interval, an image was presented for 184 ms in the gray box. 184 ms was selected because (1) it is slightly less than 200 ms, the approximate amount of time required to program and execute an eye movement, and limiting stimulus duration to less than this amount of time limits eye movements, and (2) 184 is an integer multiple of 16.66666 ms; the frame rate of the 60 Hz monitor. Selecting a stimulus duration that is an integer multiple of the frame rate reduces the likelihood that imprecise stimulus timing will occur due to the monitor’s physical limitations. Following the presentation of the stimulus, the black fixation cross re-appeared on the screen for 800 ms during which time the participants could still register a behavioral response. After this response interval, a white fixation cross appeared on the screen for 1500 ms, during which time the participants were told they could blink and move their eyes. After this blink period, the next trial began automatically. Trials progressed onward regardless of whether or not the participant made a response in the Go/NoGo task (i.e., the experiment was not selfpaced). However, this pace was chosen to be comfortable and not particularly challenging for participants. Every 100 trials, the participant was given a break.

2.5. EEG recording The EEG was recorded from 26 passive amplification ring-sintered Ag/AgCl electrodes placed in an elastic EasyCap. The electrodes were arranged geodesically.1 The electrooculogram (EOG) was recorded using two electrodes placed on the outer canthi of the left and right eyes—to monitor for horizontal eye movements – as well as one electrode placed on the sub-orbital ridge of the left eye—to monitor for blinks. Online, the EEG and EOG were referenced to the left mastoid; the EEG and vertical EOG were re-referenced offline to the average of the left and right mastoids2; the horizontal EOG was re-referenced offline as a bipolar channel. Inter-electrode impedances were kept below 2 kX (see Laszlo, Ruiz-Blondet, Khalifian, Chu, & Jin, 2014). EEG and EOG were recorded with a Brain Vision Brain Amp DC amplifier with a low pass filter at 250 Hz and a high pass filter with a 10 s time constant, and sampled at a rate of 500 Hz with an A/D resolution of 16 bits. 1 Geodesically refers to the placement of electrodes on an approximately spherical surface where each electrode is equidistantly spaced from each of its nearest neighbors. This differs from the 10–20 system, where electrodes are not all equidistant from their nearest neighbors. See Fig. 2. 2 In the N170 literature, an average reference is often used; however in the language comprehension literature averaged mastoids are more common. For this reason, we present data here with an average mastoid reference.

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

47

Fig. 2. Grand averaged ERPs elicited in response to words, objects, and ambiguous items over all 26 scalp channels for all participants. Note that, over much of the head and for much of the ERP timecourse, ERPs elicited by ambiguous items are intermediate between those elicited for words and objects, indicating that they were in fact processed ambiguously. LLOC and RLOC, the channels over which the focused N170 analysis was conducted, are boxed. In this figure, and in all subsequent figures, positive is plotted up.

Table 1 Mean lexical characteristics for word stimuli (words, outlined words, and ambiguous items). Statistics for Coltheart’s N, log bigram frequency, log syllable frequency, log orthographic frequency, word length (in letters), and number of syllables are included.

Coltheart’s N Log bigram frequency Log syllable frequency Log orthographic frequency Word length Number of syllables

Words

Outlined words

Ambiguous items

7 3.15 3.07 1.71 4.36 1.15

8.83 3.11 2.89 1.77 4.03 1.08

7.16 3.16 2.79 1.75 4.74 1.26

Each participant’s dataset was subjected to a two phase offline artifact rejection. In the first phase, ICA components were computed on high-pass filtered (0.05 Hz) data and components corresponding to blinks were removed. In the second phase, remaining artifacts less well captured by ICA (e.g., blocking, drift, horizontal eye movements) were removed with a threshold individualized to each participant. In order to be retained for data analysis, each subject’s dataset was required to retain at least 60% of trials; and >30 trials from each of the 6 categories. In the

participants retained for analysis, an average of 45.2/50 trials were retained per category. ERPs were formed for each item type (e.g., words, objects) at each electrode time-locked to stimulus onset. Included in each ERP was a 100 ms pre-stimulus baseline as well as a 998 ms post-stimulus record. ERPs were filtered with a band-pass of 0.2–20 Hz for measurement.

2.6. Statistical methods Here we adopt the linear mixed-effects regression (LMER) framework for ERP data analysis. LMER has been shown to be well-suited for the analysis of psycholinguistic experiments (e.g., Baayen, 2008). One reason for this is that it is powerful for generalizing the effects of a fixed variable (for example in the present case, item type) across one or more random variables (for example in the present case, subject) simultaneously (e.g., Baayen, Davidson, & Bates, 2008). LMER can be thought of as an extension of mixed repeated measures ANOVA – which has been a standard method for analyzing ERP data for decades (e.g., as suggested in Keselman, 1998) – from the factorial case of ANOVA to the continuous case of multiple regression.

48

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

3. Results 3.1. Reading assessments Mean performance for the vocabulary assessment was 50.3% (standard deviation, r = 19.6). The average number of items reported in the verbal fluency task was 55.2 (r = 8.3). The average proportion of correct author identifications was 33.4% (r = 3.2); the average proportion of false alarms was 2.1% (r = .90). The average proportion of correct magazine identifications was 46.2% (r = 3.8); the average proportion of false alarms was 9.0% (r = 3.2). These results indicate that the behavioral measures administered had an appropriate level of difficulty, as participants were near neither ceiling or floor on the measures provided. 3.2. EEG behavioral task While EEG was collected, participants were asked to press a button on a gamepad with their right hand whenever a picture of a non-human animal was presented (a Go/NoGo task). Participants averaged 48.08/50 (r = 1.57) hits, or 96.16%. The mean number of false alarms (a button press to a picture other than a non-human animal) was 7.87/250 (r = 5.33), or 3.2%. These statistics indicate that participants were attending appropriately to the stream of images. 3.3. ERPs Fig. 2 displays grand-averaged responses from all participants to words, objects, and ambiguous items across the entire scalp. Two central points are visible in Fig. 2. First, across the majority of scalp sites and across much of the ERP waveform, the ambiguous items are treated intermediately between words and objects. Second, the N170 is most prominent over lateral occipital channels, consistent with past work (e.g., Rossion et al., 2003). As N170 lateralization and specialization are the key issues under consideration, we began our analysis with an omnibus model aimed at examining the topographical distribution of the N170 across the scalp, and how this distribution interacted with item type. The response variable in the omnibus model is peak to peak amplitude across the P1/N170 complex; the P1 was measured from 80 to 120 ms post stimulus onset and the N170 was measured from 160 to 180 ms post stimulus onset. The P1 window was selected as 20 ms on either side of the canonical P1 peak (100 ms), and the N170 interval was selected as 10 ms on either side of the canonical N170 peak (170 ms). Minor variations in the window selected for analysis do not alter the pattern of results. Peak to peak amplitude was used instead of mean amplitude as there were differences by both item type and individual measures of reading ability on the P1 component – this is visible in Fig. 3. As the P1 precedes the N170, it is important to establish that differences on the N170 are not only a result of preceding differences on the P1, and peak to peak analysis accomplishes this. The omnibus model included subject and item as categorical factors with random intercepts. Item type (word, object, or ambiguous item), laterality (left, medial, right), and anteriority (anterior, medial, and posterior) were entered as fixed categorical factors. Verbal fluency, vocabulary, and exposure to print were included as fixed continuous factors.3 Item type was allowed to interact with anteriority and laterality, in order to examine potential topographical differences in voltage distribution in the item type effects. There was no grouping factor for participants in the model. That is, participants were not 3 An additional model including age and gender as fixed factors reveals the same pattern of results.

divided into groups on the basis of any behavioral measures into, for example, groups of poor and strong readers. Indeed, this is one of the strengths of the LMER approach to ERP data analysis – that it allows individual variability on behavioral measures to account for variance in the response variable without placing participants in monolithic groups (see Khalifian, Stites, & Laszlo, in press). As expected, item type interacted with both anteriority (F = 322, p < .0001) and laterality (F = 3.06, p = .05). In a follow up model, all predictors were kept the same except that item type was now allowed to interact with channel instead of with anteriority and laterality. This allows us to examine exactly how the interactions of item type with anteriority and laterality manifest across the scalp, explaining differences in voltage distribution for the three item types as they interact with each individual channel – this is a more fine-grained topographical analysis than that in the first omnibus model, which used the relatively large-grained topographical factors of laterality and anteriority instead of a factor of channel. This model revealed that the interaction of anteriority and item type observed in the first model is the result of the item type effects being larger over the back of the head; no reliable interactions between item type and channel were observed on the P1/N170 complex anterior of Cz. As expected, this model also revealed that the interaction between item type and laterality was driven by the item type effects being largest over the most lateral sites; indeed, the largest interactions between channel and item type over the entire head were over the right and left lateral occipital channels (RLOC: t = 10.44, p < .0001; LLOC: t = 10.1, p < .0001). This is consistent with past work indicating that N170 effects are most prominent over lateral occipital sites. The results of the omnibus models are most clearly visible in Fig. 2, which displays grand mean ERPs elicited in response to all three item types across the entire scalp. Consequently, we will focus further analysis of individual differences on the right lateral occipital and left lateral occipital electrode sites. Fig. 3 displays N170s elicited by words, objects, and word/object ambiguous items over these sites in readers with the most (top) and least (bottom) exposure to print; Fig. 4 displays each pair of item types (words and objects, objects and ambiguous items, and words and ambiguous items) again separately for readers with the most and least experience. In Fig. 4 ERPs are baselined to the P1, to more closely visualize the differences that will be reflected in the peak to peak analyses, while in Fig. 3 ERPs are baselined to a 100 ms pre-stimulus epoch, to allow more typical visualization. A striking pattern visible in both figures is that each pair of item types is differentiated less strongly in readers with less experience than in readers with more experience; this is especially pronounced over RLOC. To quantify this, we built an LMER model wherein the response variable was again peak to peak amplitude across the P1/N170 complex (P1: 80–120 ms; N170: 160– 180 ms). Subject and item were again included as categorical factors with random intercepts. Channel (RLOC or LLOC) and item type (word, object, or ambiguous item) were included as fixed categorical factors. Verbal fluency, vocabulary, and exposure to print were included as fixed continuous factors. Item type and channel were allowed to interact with each of the demographic factors (verbal fluency, vocabulary, and exposure to print). This model demonstrated that the effect of item type interacts with exposure to print (for words vs. ambiguous items, t = 3.95, p < .001; for objects vs. ambiguous items t = 2.23, p = .03, for words vs. objects t = 1.73, p = .08); this is reflected in the ERPs as smaller differences between each pair of item types in the individuals with lower exposure to print. None of the other item type by behavior interactions were significant. The effect of exposure to print was also shown in the model to interact with hemisphere (t = 4.25, p < .0001); this is observed in the ERPs as larger effects of exposure over RLOC than over LLOC (grand mean peak to peak LLOC exposure effects for words, ambiguous items, and objects are 1.53,

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

49

Fig. 3. Grand averaged ERPs elicited by words, objects, and word/object ambiguous items over left and right lateral occipital channels in readers with the most (top) and least (bottom) exposure to print. These waveforms are built on the standard pre-stimulus baseline, note that this is not the case in Fig. 4 which has been baselined on the P1 in order to visualize the peak to peak analyses.

Fig. 4. Grand averaged ERPs elicited by each pair of critical item types (words vs. objects, ambiguous items vs. objects, and ambiguous items vs. words) over the left and right lateral occipital channels. Note that these waveforms have been baselined on the P1 component instead of the pre-stimulus epoch. This is done in order to enable visualization of the peak to peak analyses more directly. For a view with the standard pre-stimulus baseline, see Fig. 3. The top row depicts ERPs elicited by readers with the most exposure to print, and the bottom row depicts ERPs elicited by readers with the least exposure to print.

4.8, and 1.7 lV, respectively, while grand mean RLOC exposure effects for words, ambiguous items, and objects are 3.5, 6.8, and 2.7 lV, respectively.) These effects are most clearly visualized in Fig. 4, which is baselined on the P1 to more closely reflect the peakto-peak analysis. The three-way interaction between exposure, hemisphere, and item type was not reliable when it was included in the same model (F = 1.29, p = .27). This null result seems to be obtained because, though the exposure effect seems to be especially large for the ambiguous items and especially over RLOC, there is also more variability across items and across subjects in the response to the ambiguous items as a class. 4. Discussion In the context of trying to better understand the function of RH OT compensation in dyslexia, we set out to examine, in detail, the representations of text computed by this region, and how these

representations might interact with individual reading ability. To these ends, we collected ERPs from individuals who varied in their verbal fluency, vocabulary, and exposure to print while they viewed words, objects, and word/object ambiguous items. We expected that weaker readers might display less specialization for these item types, treating words and objects more similarly, and that this might especially be observable in comparisons with the word/object ambiguous items. In particular, we predicted that the RH OT systems of the weaker readers might be more drawn in to the processing of these items, as a result of having a sparsely connected LH OT system that cannot as effectively inhibit any RH response to their object-like properties. We also suggested that intra-hemispheric differences between the RH and LH OT – such as their relative preference for different spatial frequency bands and their differential connectivity to the higher language comprehension system in the left hemisphere – could impact the processing of all item types, again especially the ambiguous items.

50

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

The word/object ambiguous items elicited the largest N170s in the study, especially over the RH (see Fig. 3). That N170s to these items were especially large, especially over the RH, we attribute to their novelty; there is some evidence that RH dominant N170s, as elicited by faces, are reduced with exposure (Heisz, Watter, & Shedden, 2006). Similarly, in other studies utilizing visually unfamiliar items, such as mixed-case or vertically displayed words, visual novelty can result in greater RH OT activity (Cohen, Dehaene, Vinckier, Jobert, & Montavont, 2008; Mayall, Humphreys, Mechelli, Olson, & Price, 2001). And, RH OT activation is larger to novel faces than to familiar ones (Rossion, Schiltz, Robaye, Pirenne, & Crommelinck, 2001). Perhaps more interestingly, the relationship between N170s elicited by the ambiguous items and N170s elicited by the words and objects was strikingly different in readers with more and less experience with print. Specifically, while the response to the ambiguous items was strongly differentiated from that for both words and objects in the more experienced readers, the ambiguous items, words, and objects elicited much more similar N170s in the less experienced readers. Overall, this pattern was observed more strongly over the RH than over the LH. This suggests that, especially in the RH (but also, to a lesser degree, in the LH), the OT cortex differentiates words, objects, and other similar items less strongly in less experienced readers than in more experienced readers. We propose that this pattern of de-differentiation of words, objects, and ambiguous items in less experienced readers is caused by a heightened level of processing of words (relative to objects and ambiguous items) in the RH. This pattern of results could also be observed through reduced processing of objects in the RH of less experienced readers, however, we do not believe this to be the case as our sample consisted of individuals with intact object recognition processing abilities. That exposure to print was the most potent individual factor in determining patterns of N170 responses to each item type is consistent with the prior literature demonstrating substantial expertise effects in N170 lateralization (e.g., Gauthier et al., 2003; Tanaka & Curran, 2001). We characterize this pattern of results as representing a lack of specialization for text processing in readers with less exposure to print. We propose that this lack of specialization may reflect an over-reliance on the visual object recognition system for the processing of visual word forms in these readers. That is, the privileged status of print that builds with reading experience and is associated with relatively LH biased N170 processing of text is not as deeply developed in the readers with less print exposure, with the result that they rely relatively more on the object-biased RH system for print processing, as do pre-literate children (as in Maurer et al., 2005). With text processing relying more on the RH, relatively sharp distinctions in the representations of words and objects – such as those observed here in the more experienced readers – are blurred. There are multiple possible mechanisms for this blurring. A first is that, as discussed in the Introduction, LH systems with relatively sparse connections may not learn to inhibit orthographic information from the RH as strongly as fully developed systems, resulting in increased overlap in the cortical representations of words and objects. A second is that individuals with sparsely connected LH systems rely more on RH dominant bottom-up processing for words (as described in Federmeier, 2007). Though either or both of these mechanisms may be at play during development, leading to the results observed here, the result is a lack of specialization for text processing that manifests especially in the RH OT, an area involved in object recognition. In addition to the blurring of the distinctions between words and other types of visual stimuli that may be engendered by over-reliance on the RH OT system, the representations of the words themselves seem likely to be of a lower quality when the RH is over-utilized. This is because the RH OT system is known

to be less sensitive to high spatial frequencies than the LH OT system (review in Flevaris et al., 2014), and these are precisely the frequencies needed for orthographic analysis. Consequently, the abstract orthographic representations provided by the RH OT system might not be as sharp as those that would be provided by the LH system. This may not be especially deleterious in individuals who have a normally functioning LH that can work to clean up any noisy representations produced by the RH OT system. In fact, this type of cleanup is a proposed function of the VWFA (Dehaene, Cohen, Sigman, & Vinckier, 2005) and is also a necessity in neural network models of reading dysfunction (e.g., Joanisse & Seidenberg, 1999; Plaut, McClelland, Seidenberg, & Patterson, 1996). However, in dyslexia, where the LH is disordered and therefore cannot adequately clean up the noisy RH representations, over-reliance on the RH OT system for extraction of orthographic features may become problematic. This is because any portion of the subsequent comprehension that relies on orthographic features will be disadvantaged. While the focus of analysis in this experiment was on the N170, there were observable differences in later time windows as well (seen in Fig. 3). Most notably, individuals with lower exposure to print elicited larger N400s for words than for objects or ambiguous items, while individuals with higher exposure to print elicited larger N400s for both words and ambiguous items. This result is in line with the hypothesis that more experienced readers are gleaning better quality orthographic information from the ambiguous items, and are therefore generating more meaningful semantic representations downstream, as occurs in interactive models of the N400 (e.g., Laszlo & Armstrong, 2014; Laszlo & Plaut, 2012). Here, we defined reading ability on the basis of measures of verbal fluency, vocabulary, and exposure to print – the latter being implicated in the previous literature as being related to laterality of the N170 component. We did not consider phonological ability in this group of readers, despite the strong literature suggesting that phonological ability is a critical predictor of reading ability more generally (see extensive review in NRP, 2000) and is especially impaired in dyslexia. This was in part due to the fact that the primary component of interest, the N170, is known to reflect visual processing that may vary as a result of perceptual expertise – that is, it is not thought to be related to phonological processing (although, of course, N170s are elicited during phonological processing of print, as N170s are always elicited during processing of print); this is underscored by the placement of its generators in OT cortex. Though phonological awareness is uncontroversially a primary predictor of reading performance, especially in development (extensive review in NRP, 2000), it is important to keep in mind that reading relies on other processes as well. Especially relevant to the present results, given the functional and anatomical locus of the N170 in the higher visual processing system, it is clear that properly directed visual attention is a crucial foundation of reading (e.g., Bosse et al., 2007; Franceschini, Gori, Ruffino, Pedrolli, & Facoetti, 2012; McCandliss, 2012; Vidyasagar & Pammer, 2010). With this in mind, disorder in the extraction of visual features from text – as embodied here by a lack of differentiation between words and other item types – can clearly be deleterious to reading as a whole. In drawing a link between problems of visual attention and problems with a more distal cognitive process like reading, work in the attention literature has pointed out that, in a neural network (both biological and artificial), a problem at any level can have ramifications at any other level (Vidyasagar & Pammer, 2010). This view of reading – as a process relying on the interaction of representations of many different types ranging from visual to semantic – is consonant with our suggestion here that de-differentiated or noisy orthographic representations arising from an overreliance on the RH OT system during visual word recognition can

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

have ramifications for reading as a whole, and is consistent with the ‘‘downstream’’ effects that exposure to print had for N400 processing of the ambiguous items. Nonetheless, because of the importance of phonological awareness, especially in the developmental dyslexia literature, future work following up on the present results could be addressed to determining whether phonological awareness plays a role in hemispheric dominance for print processing. 5. Conclusions In reading the fMRI literature pertaining to reading development, it is possible to get the impression that a network of left hemisphere brain regions are entirely responsible for readers’ interactions with text (e.g., ‘‘printed word and pseudoword processing implicates a left hemisphere (LH) posterior reading system. . .’’, Pugh et al., 2000, p.208, or ‘‘A range of neurobiological investigations shows a failure of left hemisphere posterior brain systems to function properly during reading in children and adults with reading disabilities.’’, Shaywitz et al., 2004, p.926). While it is certainly true that LH function – especially as pertains to phonological processing – is critical for successful reading, there is now a growing literature, much of it utilizing the ERP technique, that demonstrates crucial contributions of the RH to reading as well. Here, we sought to explore the finding that over-use of RH OT regions is associated with poorer reading ability even in non-dyslexic individuals, in particular by proposing that this relationship might signify over-use of the RH visual object recognition system during visual word recognition. In focusing on this result from the RH – the often-overlooked hemisphere in the domain of reading – we were able to identify and understand striking individual differences in how the two hemispheres are used to process text. In particular, we observed that readers with less experience with print – even those without any diagnosed learning disability – are seemingly more prone to processing text with the visual object recognition system than are readers with more experience, as evidenced by relatively dedifferentiated processing of words and objects, especially in the RH. This result provides for a novel interpretation of the finding that overactivation of the RH OT cortex is associated with decreased reading ability. Namely, it suggests that orthographic representations provided by the RH OT may be less ‘‘sharp’’ than those produced in the LH OT, and, when the LH is not functioning properly – as in dyslexia – and is therefore unable to clean these representations up as it might in normal reading, any subsequent portion of the comprehension system that relies on these representations may be disadvantaged. More generally, these results highlight the importance of understanding how the two hemispheres work together to process text, and also the importance of considering individual differences in how text is approached. Acknowledgments The authors acknowledge B.C. Armstrong, K.D. Federmeier, N. Khalifian, K.J. Kurtz, M. Stites, C. Van Petten, E.W. Wlotko, and M. Monk for insightful comments on previous versions of this manuscript. A. Bennett, K. Saturnino, A. Ceravolo, & V. Gertel were instrumental in ERP data acquisition. A. Laszlo contributed graphic design to preparation of the ambiguous items. This work was supported by awards to S.L. from NSF CAREER-1252975, NSF TWC SBE1422417, the Binghamton University Interdisciplinary Collaborative Grants program, and the Binghamton University Health Sciences Transdisciplinary Area of Excellence. E.S. was supported by a Binghamton University Provost’s Fellowship.

51

Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bandl.2015.03. 009. References Baayen, R. H. (2008). Analyzing linguistic data. A practical introduction to statistics using R. Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59(4), 390–412. Boetsch, E. A., Green, P. A., & Pennington, B. F. (1996). Psychosocial correlates of dyslexia across the life span. Development and Psychopathology, 8, 539–562. Bosse, M. L., Tainturier, M. J., & Valdois, S. (2007). Developmental dyslexia: The visual attention span deficit hypothesis. Cognition, 104(2), 198–230. Brysbaert, M., & New, B. (2009). Moving beyond Kucera and Francis: A critical evaluation of current word frequency norms and the introduction of a new and improved word frequency measure for American English. Behavior Research Methods, 41(4), 977–990. Cai, Q., Paulignan, Y., Brysbaert, M., Ibarrola, D., & Nazir, T. A. (2010). The left ventral occipito-temporal response to words depends on language lateralization but not on visual familiarity. Cerebral Cortex, 20(5), 1153–1163. Cohen, L., & Dehaene, S. (2004). Specialization within the ventral stream: The case for the visual word form area. Neuroimage, 22(1), 466–476. Cohen, L., Dehaene, S., Vinckier, F., Jobert, A., & Montavont, A. (2008). Reading normal and degraded words: Contribution of the dorsal and ventral visual pathways. Neuroimage, 40(1), 353–366. Cohen, L., Martinaud, O., Lemer, C., Lehericy, S., Samson, Y., Obadia, M., et al. (2003). Visual word recognition in the left and right hemispheres: Anatomical and functional correlates of peripheral alexias. Cerebral Cortex, 13(12), 1313–1333. Cohen, M. J., Morgan, A. M., Vaughn, M., Riccio, C. A., & Hall, J. (1999). Verbal fluency in children: Developmental issues and differential validity in distinguishing children with attention-deficit hyperactivity disorder and two subtypes of dyslexia. Archives of Clinical Neuropsychology, 14(5), 433–443. Coulson, S., & Williams, R. F. (2005). Hemispheric asymmetries and joke comprehension. Neuropsychologia, 43(1), 128–141. Cunningham, A. E., & Stanovich, K. E. (1991). Tracking the unique effects of print exposure in children: Associations with vocabulary, general knowledge, and spelling. Journal of Educational Psychology, 83(2), 264. Dehaene, S., & Cohen, L. (2011). The unique role of the visual word form area in reading. Trends in Cognitive Sciences, 15(6), 254–262. Dehaene, S., Cohen, L., Sigman, M., & Vinckier, F. (2005). The neural code for written words: A proposal. Trends in Cognitive Sciences, 9(7), 335–341. Federmeier, K. D. (2007). Thinking ahead: The role and roots of prediction in language comprehension. Psychophysiology, 44(4), 491–505. Federmeier, K. D., & Kutas, M. (1999). A rose by any other name: Long-term memory structure and sentence processing. Journal of Memory and Language, 41(4), 469–495. Flevaris, A. V., Martinez, A., & Hillyard, S. (2014). Attending to global versus local stimulus features modulates neural processing of low versus high spatial frequencies: an analysis with event-related brain potentials. Frontiers in psychology, 5. Franceschini, S., Gori, S., Ruffino, M., Pedrolli, K., & Facoetti, A. (2012). A causal link between visual spatial attention and reading acquisition. Current Biology, 22(9), 814–819. Gauthier, I., Curran, T., Curby, K. M., & Collins, D. (2003). Perceptual interference supports a non-modular account of face processing. Nature Neuroscience, 6(4), 428–432. Grainger, J., & Holcomb, P. J. (2009). Watching the word go by: On the time-course of component processing in visual word recognition. Language and Linguistics Compass, 3, 128–156. Grill-Spector, K., Kourtzi, Z., & Kanwisher, N. (2001). The lateral occipital complex and its role in object recognition. Vision Research, 41(10), 1409–1422. Harm, M. W., & Seidenberg, M. S. (2004). Computing the meanings of words in reading: Cooperative division of labor between visual and phonological processes. Psychological Review, 111(3), 662. Haxby, J. V., Grady, C. L., Horwitz, B., Ungerleider, L. G., Mishkin, M., Carson, R. E., et al. (1991). Dissociation of object and spatial visual processing pathways in human extrastriate cortex. Proceedings of the National Academy of Sciences, 88(5), 1621–1625. Heisz, J. J., Watter, S., & Shedden, J. M. (2006). Automatic face identity encoding at the N170. Vision Research, 46(28), 4604–4614. Hoeft, F., Meyler, A., Hernandez, A., Juel, C., Taylor-Hill, H., Martindale, J. L., et al. (2007). Functional and morphometric brain dissociation between dyslexia and reading ability. Proceedings of the National Academy of Sciences, 104(10), 4234–4239. Joanisse, M. F., & Seidenberg, M. S. (1999). Impairments in verb morphology after brain injury: A connectionist model. Proceedings of the National Academy of Sciences, 96(13), 7592–7597. Kandhadai, P., & Federmeier, K. D. (2008). Summing it up: Semantic activation processes in the two hemispheres as revealed by event-related potentials. Brain Research, 1233, 146–159.

52

S. Laszlo, E. Sacchi / Brain & Language 145–146 (2015) 42–52

Kandhadai, P., & Federmeier, K. D. (2010). Automatic and controlled aspects of lexical associative processing in the two cerebral hemispheres. Psychophysiology, 47(4), 774–785. Keller, T. A., & Just, M. A. (2009). Altering cortical connectivity: Remediationinduced changes in the white matter of poor readers. Neuron, 64(5), 624–631. Keselman, H. J. (1998). Testing treatment effects in repeated measures designs: An update for psychophysiological researchers. Psychophysiology, 35(04), 470–478. Khalifian, N., Stites, M., & Laszlo, S. (in press). Relationships between event-related potentials and behavioral and scholastic measures of reading ability: A largescale, cross-sectional study. Developmental Science. Kitterle, F. L., Christman, S., & Hellige, J. B. (1990). Hemispheric differences are found in the identification, but not the detection, of low versus high spatial frequencies. Perception & Psychophysics, 48(4), 297–306. Laszlo, S., & Armstrong, B. C. (2014). PSPs and ERPs: Applying the dynamics of postsynaptic potentials to individual units in simulation of ERP reading data. Brain and Language, 132, 22–27. Laszlo, S., & Federmeier, K. D. (2014). Never seem to find the time: Evaluating the physiological time course of visual word recognition with regression analysis of single-item event-related potentials. Language, Cognition and Neuroscience, 29, 642–661. Laszlo, S., & Plaut, D. C. (2012). A neurally plausible parallel distributed processing model of event-related potential word reading data. Brain and Language, 120, 271–281. Laszlo, S., Ruiz-Blondet, M., Khalifian, N., Chu, F., & Jin, Z. (2014). A direct comparison of active and passive amplification electrodes in the same amplifier system. Journal of Neuroscience Methods, 235, 298–307. Lindgren, S. D., De Renzi, E., & Richman, L. C. (1985). Cross-national comparisons of developmental dyslexia in Italy and the United States. Child Development, 1404–1417. Liu, Y., Dunlap, S., Fiez, J., & Perfetti, C. (2007). Evidence for neural accommodation to a writing system following learning. Human Brain Mapping, 28(11), 1223–1234. Mahé, G., Bonnefond, A., Gavens, N., Dufour, A., & Doignon-Camus, N. (2012). Impaired visual expertise for print in French adults with dyslexia as shown by N170 tuning. Neuropsychologia, 50(14), 3200–3206. Maurer, U., Brem, S., Bucher, K., & Brandeis, D. (2005). Emerging neurophysiological specialization for letter strings. Journal of Cognitive Neuroscience, 17(10), 1532–1552. Maurer, U., & McCandliss, B. D. (2008). The development of visual expertise for words: The contribution of electrophysiology. Single Word Reading: Behavioral and Biological Perspectives, 43–63. Mayall, K., Humphreys, G., Mechelli, A., Olson, A., & Price, C. (2001). The effects of case mixing on word recognition: Evidence from a PET study. Journal of Cognitive Neuroscience, 13(6), 844–853. McCandliss, B. D. (2012). Helping dyslexic children attend to letters within visual word forms. Proceedings of the National Academy of Sciences, 109(28), 11064–11065. Medler, D. A., & Binder, J. R. (2005). MCWord: An on-line orthographic database of the English language. . Mei, L., Xue, G., Lu, Z. L., He, Q., Zhang, M., Wei, M., et al. (2014). Artificial language training reveals the neural substrates underlying addressed and assembled phonologies. PLoS One, 9(3), e93548. National Reading Panel (US), National Institute of Child Health, & Human Development (US) (2000). Report of the National Reading Panel: Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction: Reports of the subgroups. National Institute of Child Health and Human Development, National Institutes of Health. Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9(1), 97–113. Paulesu, E., Démonet, J. F., Fazio, F., McCrory, E., Chanoine, V., Brunswick, N., et al. (2001). Dyslexia: Cultural diversity and biological unity. Science, 291(5511), 2165–2167. Plaut, D. C., McClelland, J. L., Seidenberg, M. S., & Patterson, K. (1996). Understanding normal and impaired word reading: Computational principles in quasi-regular domains. Psychological Review, 103, 56–115. Price, C. J., & Devlin, J. T. (2003). The myth of the visual word form area. Neuroimage, 19(3), 473–481. Price, C. J., & Devlin, J. T. (2011). The interactive account of ventral occipitotemporal contributions to reading. Trends in Cognitive Sciences, 15(6), 246–253. Pugh, K. R., Frost, S. J., Sandak, R., Landi, N., Moore, D., Della Porta, G., et al. (2010). Mapping the word reading circuitry in skilled and disabled readers. The Neural Basis of Reading, 281–305. Pugh, K. R., Mencl, W. E., Jenner, A. R., Katz, L., Frost, S. J., Lee, J. R., et al. (2000). Functional neuroimaging studies of reading and reading disability(developmental dyslexia). Mental Retardation and Developmental Disabilities Research Reviews, 6(3), 207–213.

Rashotte, C. A., & Torgesen, J. K. (1985). Repeated reading and reading fluency in learning disabled children. Reading Research Quarterly, 180–188. Reiter, A., Tucha, O., & Lange, K. W. (2005). Executive functions in children with dyslexia. Dyslexia, 11(2), 116–131. Richards, T. L., & Berninger, V. W. (2008). Abnormal fMRI connectivity in children with dyslexia during a phoneme task: Before but not after treatment. Journal of Neurolinguistics, 21(4), 294–304. Rossion, B., & Caharel, S. (2011). ERP evidence for the speed of face categorization in the human brain: Disentangling the contribution of low-level visual cues from face perception. Vision Research, 51(12), 1297–1311. Rossion, B., Collins, D., Goffaux, V., & Curran, T. (2007). Long-term expertise with artificial objects increases visual competition with early face categorization processes. Journal of Cognitive Neuroscience, 19(3), 543–555. Rossion, B., & Jacques, C. (2012). The N170: Understanding the time course of face perception in the human brain. In S. J. Luck & E. S. Kappenman (Eds.), The Oxford handbook of event-related potential components (pp. 115–141). New York, NY: Oxford University Press. Rossion, B., 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(3), 1609–1624. Rossion, B., Schiltz, C., Robaye, L., Pirenne, D., & Crommelinck, M. (2001). How does the brain discriminate familiar and unfamiliar faces?: A PET study of face categorical perception. Journal of Cognitive Neuroscience, 13(7), 1019–1034. Seghier, M. L., & Price, C. J. (2011). Explaining left lateralization for words in the ventral occipitotemporal cortex. The Journal of Neuroscience, 31(41), 14745–14753. Shaywitz, S. E., & Shaywitz, B. A. (2005). Dyslexia (specific reading disability). Biological Psychiatry, 57(11), 1301–1309. Shaywitz, S. E., & Shaywitz, B. A. (2008). Paying attention to reading: The neurobiology of reading and dyslexia. Development and psychopathology, 20(4), 1329–1349. Shaywitz, B. A., Shaywitz, S. E., Blachman, B. A., Pugh, K. R., Fulbright, R. K., et al. (2004). Development of left occipitotemporal systems for skilled reading in children after a phonologically-based intervention. Biological Psychiatry, 55(9), 926–933. Shaywitz, S. E., Shaywitz, B. A., Pugh, K. R., Fulbright, R. K., Constable, R. T., Mencl, W. E., et al. (1998). Functional disruption in the organization of the brain for reading in dyslexia. Proceedings of the National Academy of Sciences, 95(5), 2636–2641. Spiridon, M., & Kanwisher, N. (2002). How distributed is visual category information in human occipito-temporal cortex? An fMRI study. Neuron, 35(6), 1157–1165. Stanovich, K. E., & Cunningham, A. E. (1992). Studying the consequences of literacy within a literate society: The cognitive correlates of print exposure. Memory & Cognition, 20(1), 51–68. Stanovich, K. E., & West, R. F. (1989). Exposure to print and orthographic processing. Reading Research Quarterly, 402–433. Stanovich, K. E., West, R. F., & Cunningham, A. E. (1991). Beyond phonological processes: Print exposure and orthographic processing. Phonological Processes in Literacy: A Tribute to Isabelle Y. Liberman, 219–235. Steinbrink, C., Vogt, K., Kastrup, A., Müller, H. P., Juengling, F. D., Kassubek, J., et al. (2008). The contribution of white and gray matter differences to developmental dyslexia: Insights from DTI and VBM at 3.0 T. Neuropsychologia, 46(13), 3170–3178. Tanaka, J. W., & Curran, T. (2001). A neural basis for expert object recognition. Psychological Science, 12(1), 43–47. Tarr, M. J., & Vuong, Q. C. (2002). Visual object recognition. Stevens’ Handbook of Experimental Psychology. Temple, E., Deutsch, G. K., Poldrack, R. A., Miller, S. L., Tallal, P., Merzenich, M. M., et al. (2003). Neural deficits in children with dyslexia ameliorated by behavioral remediation: Evidence from functional MRI. Proceedings of the National Academy of Sciences, 100(5), 2860–2865. Van der Mark, S., Klaver, P., Bucher, K., Maurer, U., Schulz, E., Brem, S., et al. (2011). The left occipitotemporal system in reading: Disruption of focal fMRI connectivity to left inferior frontal and inferior parietal language areas in children with dyslexia. Neuroimage, 54(3), 2426–2436. Vidyasagar, T. R., & Pammer, K. (2010). Dyslexia: A deficit in visuo-spatial attention, not in phonological processing. Trends in Cognitive Sciences, 14(2), 57–63. West, R. F., & Stanovich, K. E. (1991). The incidental acquisition of information from reading. Psychological Science, 2(5), 325–330. Wlotko, E. W., & Federmeier, K. D. (2007). Finding the right word: Hemispheric asymmetries in the use of sentence context information. Neuropsychologia, 45(13), 3001–3014. Yoncheva, Y. N., Blau, V. C., Maurer, U., & McCandliss, B. D. (2010). Attentional focus during learning impacts N170 ERP responses to an artificial script. Developmental Neuropsychology, 35(4), 423–445.