Content area
Language processing is often an area of difficulty in Autism Spectrum Disorder (ASD). Semantic processing—the ability to add meaning to a stimulus—is thought to be especially affected in ASD. However, the neurological origin of these deficits, both structurally and temporally, have yet to be discovered. To further previous behavioral findings on language differences in ASD, the present study used an implicit semantic priming paradigm and electroencephalography (EEG) to compare the level of theta coherence throughout semantic processing, between typically developing (TD) and ASD participants. Theta coherence is an indication of synchronous EEG oscillations and was of particular interest due to its previous links with semantic processing. Theta coherence was analyzed in response to semantically related or unrelated pairs of words and pictures across bilateral short, medium, and long electrode connections. We found significant results across a variety of conditions, but most notably, we observed reduced coherence for language stimuli in the ASD group at a left fronto-parietal connection from 100 to 300 ms. This replicates previous findings of underconnectivity in left fronto-parietal language networks in ASD. Critically, the early time window of this underconnectivity, from 100 to 300 ms, suggests that impaired semantic processing of language in ASD may arise during pre-semantic processing, during the initial communication between lower-level linguistic processing and higher-level semantic processing. Our results suggest that language processing functions are unique in ASD compared to TD, and that subjects with ASD might rely on a temporally different language processing loop altogether.
Introduction
Successful language processing involves the careful and well-timed coordination of component processes, achieved by the synchronous activity of brain regions at each moment of processing. For typically developing (TD) individuals, this process occurs relatively automatically and with apparent ease. But should aspects of this process go awry, even small deviations in the execution of component processes could have significant effects on the final output.
Language in autism
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by deficits in social communication and restricted and repetitive behaviors (American Psychiatric Association 2013). Although no longer a diagnostic criterion, difficulties with various aspects of language processing are common (Tager‐Flusberg et al. 2005). However, language is not universally affected in ASD; some aspects of language processing pose more challenges than others.
Lower-level linguistic processing in ASD
Word decoding is a lower-level linguistic skill that involves the ability to identify a word and its correct pronunciation, independent from its meaning (Swank and Catts 1994). When looking at the ability to decode words and understand their meanings both in isolation and in different contexts, participants with ASD have shown intact word decoding abilities (Frith and Snowling 1983; Whitehouse and Harris 1984; Huemer and Mann 2010; Norbury and Nation 2011). In fact, findings of “hyperlexia”, or superior word reading skills compared to neurotypical peers, occur relatively frequently among individuals with ASD (Ostrolenk et al. 2017). Neuroimaging studies of word reading in subjects with ASD, including those with hyperlexia, have suggested that word reading is also associated with greater activation of visual perception and form recognition brain regions compared to TD populations (Turkeltaub et al. 2004; Samson et al. 2012). Overall, these findings suggest that lower-order linguistic processing is intact in ASD, and may be related to enhanced visual processing, as has been proposed by other models of autism such as the Enhanced Perceptual Functioning Model (Mottron et al. 2006).
Higher-level linguistic processing in ASD
Semantic processing refers to the ability to receive a stimulus—such as a spoken or written word, picture, or sound—and apply meaning to it based on previously stored knowledge (Kelley et al. 2006; Yang and Small 2015). For linguistic stimuli,1 difficulties with semantic processing are common among individuals with ASD (Tager‐Flusberg 1991; Kamio and Toichi 2000; Kamio et al. 2007; Coderre et al. 2017). Even people considered to be in the “optimal outcome” stage of ASD, meaning they were diagnosed at a very young age but no longer meet diagnostic criteria due to early intervention or developmental changes, retain atypicalities in their semantic processing (Kelley et al. 2006). These difficulties with semantic processing, despite intact or even superior word reading capabilities, result in a characteristic asymmetry in language abilities among individuals with ASD (Tager-Flusberg et al. 2005; Ostrolenk et al. 2017).
Higher-level language functions such as semantic processing often require the assimilation of different types of information (e.g. linguistic and semantic, linguistic and social) and thus draw heavily on semantic integration: the ability to integrate the meanings of many separate pieces of information to arrive at a holistic understanding. Given the difficulties observed with these higher-level functions in ASD, some theories propose that ASD is primarily an integration disorder (Frith 1989; Brock et al. 2002). For example, the theory of weak central coherence (Frith 1989) posits that individuals with ASD favor a strategy of local rather than global processing, exhibiting a piecemeal processing style and a reduced tendency to integrate information with its larger context.
Impaired semantic processing in ASD has been documented by a number of studies using both behavioral (Frith and Snowling 1983; Happé 1997; Jolliffe and Baron-Cohen 2000; López and Leekam 2003) and neural measures (Verbaten et al. 1991; Strandburg et al. 1993; Dunn et al. 1999; Valdizán et al. 2003; Just et al. 2004; Dunn and Bates 2005; Harris et al. 2006; Gaffrey et al. 2007; Braeutigam et al. 2008; McCleery et al. 2010; Pijnacker et al. 2010; Fishman et al. 2011; Ribiero et al. 2013; Coderre et al. 2017; DiStefano et al. 2019; Manfredi et al. 2020). Event-related potentials (ERPs) are time-locked brain responses derived from the electroencephalogram (EEG) that reflect cognitive functioning following the presentation of a stimulus. The N400 ERP component is specifically thought to reflect semantic processing, and is named as such because the peak occurs 400 milliseconds (ms) after the presentation of a stimulus (Lau et al. 2008; Kutas and Federmeier 2011; Kulakova and Nieuwland 2016). In the TD population, an N400 effect is elicited when a participant recognizes a disruption in semantics, which typically occurs when participants compare congruent and incongruent words, or related and unrelated words (Holcomb 1993; Lau et al. 2008; Kutas and Federmeier 2011). In general, the easier it is to access a word’s semantic representation or integrate that semantic representation with a prior semantic context, the smaller the N400 amplitude.
Despite it being an established ERP component, there is ample discussion surrounding the exact onset and end time of the N400. Some researchers argue that the N400 begins immediately following early lexical access processes, around 250 ms after stimulus presentation (Kutas and Federmeier 2009; Renoult et al. 2012). This onset would immediately follow the N170 ERP component, which reflects the identification of letter strings and peaks 170 ms after the presentation of a word (Sereno and Rayner 2003). Other researchers argue that the N400 has a later onset, closer to 300–350 ms (Friederici et al. 2004; Kutas and Federmeier 2009), while still others argue that the N400 encompasses higher-order processes such as syntactic processing (Brouwer et al. 2017). Despite the varying opinions on the neural processes represented by the N400, one idea has remained constant: the order of ERP latencies mirrors a hierarchy of complexity in the neural processes they portray. It is agreed upon that earlier neural activity represents lower-level neural processing compared to later activity peaks due to the need for basic language coding processes, such as lexical access, to occur before information can be synthesized into a more complex understanding.
Previous work has shown that individuals with ASD have reduced or absent N400 effects (i.e., the modulation of N400 amplitude by integration difficulty) in response to language, suggesting difficulties with semantic processing and integration (Verbaten et al. 1991; Strandburg et al. 1993; Dunn et al. 1999; Dunn and Bates 2005; Braeutigam et al. 2008; McCleery et al. 2010; Pijnacker et al. 2010; Fishman et al. 2011; Ribiero et al. 2013; Coderre et al. 2017; DiStefano et al. 2019; Manfredi et al. 2020). For example, when performing a semantic priming task in which participants were presented with pairs of written and spoken words that were either matched or mismatched, McCleery et al. (2010) observed an N400 effect for mismatching picture/word visual stimuli in TD children, but not in children with ASD. This suggests that children with ASD were not as sensitive to the semantic relationship between the linguistic and visual stimuli, and that the semantic response to single words is impaired in ASD. These findings were specific to the verbal domain only (McCleery et al., 2010). Some other studies have found that although the amplitude of the N400 did not differ in ASD children compared to TD children, ASD participants showed an increase in N400 latency compared to controls (Valdizán et al. 2003; Mendez et al. 2009; DiStefano et al. 2019). DiStefano et al. (2019) used a picture-word matching paradigm where TD, verbal ASD, and minimally verbal ASD participants aged 5–11 years old were shown pairs of either linguistic or non-linguistic basic nouns. They, similar to Valdizán et al. (2003), found a greater latency of the N400 in both ASD groups (verbal and minimally verbal).
One study comparing TD and ASD responses to pictures primed with congruent or incongruent spoken sentences or popular musical excerpts found comparable levels of accuracy between groups (when asked whether the stimuli were congruent or incongruent) but larger N400s in the TD group than the ASD group in response to the sentence primes (Ribiero et al. 2013). The ASD group however, showed a larger late positive potential (LPP) than the TD group (Ribiero et al. 2013), thought to represent later stages of semantic integration, stimuli reanalysis, and memory formation (Caplan et al. 2009). This decoupling of the N400 and LPP and the reversed direction of group effects suggest that ASD participants process semantic stimuli differently than the TD group.
As these studies show, the literature surrounding the N400 in ASD is very contrasted. Pijnacker et al. (2010) found contradictory results to those of Valdizán et al. (2003) and DiStefano et al. (2019) when measuring context sensitivity and context-specific reasoning between ASD and control groups. They tested two conditions: reading semantically congruent or incongruent sentences, or processing congruent or incongruent reasoning problems. Their ASD participants showed no N400 effect across either condition, whereas the control group showed N400 effects in both. The control group also showed a later positive peak during the sentence context, and sustained negative activity during the reasoning context, while the ASD participants showed only a late positive peak under both conditions. This peak was larger in response to semantically anomalous sentences than to congruent ones, but no sustained negativity was present. Other work has linked the reduced N400 and the later positivity, suggesting that the lack of an initial N400 effect could represent a lack of early semantic processing in ASD, requiring participants to double back and reinterpret the stimulus later on (Groen et al. 2008; Nijhof et al. 2018), which is reflected by a later positivity.
Our own previous work (Coderre et al. 2017; O’Rourke & Coderre, 2021) has found similar N400 effects between ASD and TD groups, but more subtle differences upon further investigation. In Coderre et al. (2017), which used a similar semantic priming paradigm as employed here, both groups showed N400 effects in response to word stimuli, contrary to prior literature. However, differences in N400 latency and topography in word conditions suggested different lexico-semantic processing mechanisms: an expectancy-based strategy for the TD group but a controlled post-lexical integration strategy for the ASD group. In O’Rourke & Coderre (2021), which uses the same dataset as the current study, we found no significant differences in N400 effects between TD and ASD groups for word stimuli, but found stronger effects when splitting participants by levels of autistic traits. Participants with higher levels of autistic traits showed increased difficulty with semantic processing of language (i.e., smaller N400s) than those with lower levels of autistic traits. Regardless of whether the N400’s amplitude, latency, or duration is altered in ASD, the previously mentioned studies show that it differs, in many cases, to that recorded in TD participants.
Functional magnetic resonance imaging (fMRI) research has demonstrated that important brain areas for semantic integration, such as the left inferior frontal gyrus (LIFG) (Hagoort 2005), show reduced activation during language processing in individuals with ASD compared to TD controls (Just et al. 2004; Harris et al. 2006; Gaffrey et al. 2008). For example, Just et al. (2004) performed a sentence comprehension task during fMRI and reported that individuals with ASD showed greater activation in Wernicke's area, the left superior temporal gyrus, but less activation in Broca's area, the left inferior frontal gyrus, compared to controls. They interpreted this as suggesting that individuals with ASD had more extensive processing of the meaning of single words, as evidenced by the greater Wernicke's activation, but had difficulties with integrating the words and processing the meaning of complex sentences, as evidenced by the reduced Broca’s activation. Overall, these neuroimaging data, from both ERP and fMRI studies, suggest difficulties with semantic integration in ASD.
Non-linguistic semantic processing
Taken together, the current evidence suggests that lower-level linguistic functions like word decoding are intact in individuals with ASD, whereas higher-level linguistic functions like semantic processing and integration are impaired.
One possibility for the discrepancy between lower- and higher-level linguistic functions could be that the semantic system is fundamentally impaired in ASD. However, there is evidence that semantic processing in individuals with ASD is often intact when non-linguistic stimuli are used (Kamio and Toichi 2000; Sahyoun et al. 2009; McCleery et al. 2010). For example, in a priming paradigm, Kamio & Toichi (2000) compared the integration of pairs of pictures and words versus pairs of words in children with ASD. Whereas TD children performed similarly on both tasks, children with ASD performed better on the picture-word task, suggesting an advantage for visuo-semantic processing over lexico-semantic processing. Similarly, McCleery et al. (2010) found that children with ASD showed no N400 effect in response to a linguistic condition (pairs of matching or mismatching pictures and words), but they did show an N400 effect of similar magnitude to TD children for a non-linguistic condition (pairs of pictures and environmental sounds). They interpreted this finding as indicative of intact semantic processing for non-linguistic stimuli but impaired semantic processing for linguistic stimuli. In two fMRI studies, Sahyoun and colleagues (2009) found that individuals with ASD favored visual processing over lexico-semantic processing (Sahyoun et al. 2009) and relied more on visual processing areas during lexico-semantic processing (Sahyoun et al. 2010a) compared to TD individuals.
However, many of these previous studies have included cross-modal stimuli when examining semantic processing in ASD, such as pictures paired with words (Kamio and Toichi 2000; McCleery et al. 2010). Yet the presence of words in a “non-linguistic” condition, or the presence of pictures in a “linguistic” condition, may modulate semantic processing mechanisms if individuals with ASD truly do experience a modality-specific deficit. For this reason, in prior studies we have developed a within-modality semantic priming paradigm that consists of picture-picture or word-word pairs (Coderre et al. 2017; Coderre, 2018). Being able to isolate linguistic and non-linguistic semantic processing using within-modality stimuli allows for a cleaner comparison between modalities to determine if individuals with ASD truly do experience a “sparing” in non-linguistic modalities (Coderre 2020).
In a recent ERP study by our group, we employed this same semantic priming paradigm, in which participants viewed semantically related or unrelated pairs of pictures or words, to compare implicit lexico-semantic and visuo-semantic processing between individuals with ASD and TD individuals (O’Rourke and Coderre 2021). We found a larger N400 effect in response to pictures for the ASD group compared to the TD group, suggesting that visuo-semantic processing may be enhanced in individuals with ASD, in line with the Enhanced Perceptual Functioning Model (Mottron et al. 2006). In response to words, we found that individuals with more severe autistic traits showed smaller N400 effects, reflecting a reduced sensitivity to semantic relatedness in linguistic stimuli. Manfredi et al. (2020) expanded on our previous results using a similar paradigm but with visual, non-linguistic (picture) and auditory linguistic (instead of visual) stimuli. This explicit design asked TD and ASD participants whether each sentence was understandable after each stimulus. Their younger sample (aged 9–15) showed an attenuated N400 effect to incongruent words in the TD group, but not in the ASD group (Manfredi et al. 2020). Incongruent pictures, however, elicited an increased N400 in both groups (Manfredi et al. 2020). Overall, these findings—both from our group and others—suggest that semantic processing is not impaired when visual stimuli are used, but that challenges selectively arise during semantic processing of linguistic stimuli.
If semantic processing is intact in some domains, such as visual processing, this suggests that there may be a language-specific deficit in semantic processing in ASD, such that semantic processing poses more challenges when linguistic stimuli are used to access semantics. The field of psychology has long debated whether semantic processing relies on a single “amodal hub” or is separated based on specific input modalities. Neuroscience evidence has seemed to support the idea that lexico-semantic processing (i.e. semantic processing of linguistic stimuli) and visuo-semantic processing (i.e. semantic processing of visual stimuli) share parts of a core semantic network, primarily involving the anterior temporal lobe (ATL). Jackson et al. (2016) found the ATL, specifically its lateral regions (ventral ATL), to be connected to a single semantic network in both resting and task completion states. The network also included regions involved in semantic cognition such as the bilateral ATL, inferior frontal gyrus (IFG), medial pre-frontal cortex (mPFC), angular gyrus (AG) and posterior medial temporal gyrus (pMTG) (Jackson et al. 2016). The vATL’s similar connectivity pattern across active task and resting states suggests that it is a core network responsible for multimodal semantic cognition. While the central semantic processing network is considered amodal, different modalities access this network via different connections. The ATL amodal hub interacts with modality-specific regions to allow for semantic processing in each modality (Jackson et al. 2016). Lexico-semantic processing tends to activate left prefrontal and parietal areas while visuo-semantic processing activates temporal, parietal, and occipital areas (Vandenberghe et al. 1996; Cabeza and Nyberg 2000; Rossion et al. 2000; Price 2010). Here, we will propose that it is these modality-specific connections to the amodal semantic system that may be selectively affected for language in individuals with ASD.
Neural connectivity in ASD
To summarize, we have seen that individuals with ASD experience intact lower-level linguistic processing (e.g. word decoding), suggesting that the language system is not fundamentally compromised in ASD. At the same time, individuals with ASD experience impaired semantic processing in language, suggesting impairments in higher-level linguistic processing. However, these semantic processing deficits are not universal: semantic processing of non-linguistic stimuli is often intact. This suggests that the semantic system is not fundamentally compromised in ASD either.
If lower-level, language-specific functions are intact, and higher-level non-linguistic functions are intact, then where is the miscommunication occurring to create higher-level linguistic deficits? Since the literature shows the beginning and end of the system to be fundamentally functional, perhaps it is connections between lower-level and higher- level processes that are affected (see Fig. 1). We propose that the structural and/or functional neural connections between brain areas involved in earlier stages such as word processing and later stages such as semantic processing may be selectively disrupted in individuals with ASD. That is, difficulties with semantic processing in language in ASD may not stem from difficulties with either the linguistic or the semantic systems, but the connections between the two.
Fig. 1 [Images not available. See PDF.]
Theoretical model of lexico-semantic difficulties in ASD. a) One possibility is that the language processing system is impaired; however, findings of intact lower-level linguistic processing rule out this scenario. b) Another option is that the semantic processing system is impaired; however, findings of intact visuo-semantic processing rule out this scenario. c) A third option is that the connections between the language and semantic systems are impaired, which we propose to test in the current study
This proposal is in line with prior neuroimaging studies investigating neural connectivity in individuals with ASD. MRI data has indicated that, compared to TD individuals, individuals with ASD show atypical structural connectivity (Sahyoun et al. 2010a; Zikopoulos and Barbas 2013) across various regions of the brain. In particular, the arcuate fasciculus, a white matter tract connecting Broca’s area in the frontal lobe to Wernicke’s area in the parietal lobe, shows differences in volume, diffusivity, and functional connectivity in ASD (Ben Bashat et al. 2007; Fletcher et al. 2010; Wan et al. 2012; Roberts et al. 2014; Moseley et al. 2016). The arcuate fasciculus is part of the larger semantic network, but its left hemisphere tract has been found to play a critical role in language production and comprehension (Catani and Mesulam 2008; Lebel and Beaulieu 2009; Yeatman et al. 2011; Wan et al. 2012; Lopez-Barroso et al. 2013; Joseph et al. 2014; Roberts et al. 2014; Moseley et al. 2016). In TD participants, the arcuate fasciculus is generally structurally and functionally left-lateralized, emphasizing its role in the left-lateralized language network (Lebel and Beaulieu 2009; Eichert et al. 2019); however, in ASD, this tract is more bilateral, meaning that language specialization is not as evident (Fletcher et al. 2010; Wan et al. 2012; Moseley et al. 2016). Underconnectivity in the arcuate fasciculus may thus suggest a structural basis for semantic processing difficulties in ASD.
In addition to differences in structural connectivity, abnormal patterns of functional connectivity have also been documented in ASD. Overall, a pattern of long-range underconnectivity and short-range overconnectivity is documented in ASD (Just et al. 2004; Cherkassky et al. 2006; Kana et al. 2006; Murias et al. 2007; Coben et al. 2008; Jones et al. 2010; Barttfeld et al. 2011; Anderson et al. 2011; Catarino et al. 2013; Parellada et al. 2014; Baribeau and Anagnostou 2015; Hull et al. 2017). Some have suggested that this neural underconnectivity may give rise to the cognitive, perceptual, and motor difficulties characterizing ASD (Just et al. 2004, 2007). For instance, the temporal binding deficit hypothesis (Brock et al. 2002) proposes that hypocoupling of neural networks in the brain impairs the synchronous firing of neurons that underlies the processing and integration of information (‘temporal binding’), leading to behavioral manifestations of weak central coherence (Frith 1989). This atypical connectivity likely plays a significant role in the semantic processing impairments noted in ASD. As we mentioned above, language processing requires the careful and precise synchronization of numerous brain areas at each level of processing; therefore, any alterations in this synchronization and communication, however slight, likely have significant impacts on the end result.
Many studies have found this underconnectivity to be particularly prominent in left fronto-parietal connections. Apart from the arcuate fasciculus showing decreased cortical volume in ASD, a hypothesized result of underconnectivity, decreased cortical volume has also been noted in the inferior frontal gyrus in individuals with ASD (Joseph et al. 2014). During a working memory task, Koshino et al. (2005) found left fronto-parietal activity to be lower in ASD participants, despite both ASD and TD groups having comparable right hemisphere activity levels (Koshino et al. 2005). When asked to comprehend high-imagery sentences, which verbally described the visual characteristics of objects, ASD participants showed less functionally synchronous fMRI activity in fronto-parietal areas than the TD group (Kana et al., 2006). Kana et al. (2006) also found the volume of the anterior corpus callosum, in the fronto-parietal region, to be slightly smaller in ASD than in TD. This left hemisphere fronto-parietal underconnectivity in ASD has also been found during sentence comprehension tasks (Just 2004), linking these structural differences to language deficits in ASD.
In summary, this evidence of neural underconnectivity in individuals with ASD, particularly in left fronto-parietal connections that are important for language, lends support to the possibility that the language-specific semantic processing deficits in ASD arise from impaired connections between the language and semantic systems, while connections between the visual and semantic systems are not affected (Fig. 1c). As visuo-semantic processing typically involves temporo- parietal areas of the brain (Vandenberghe et al. 1996; Martin and Chao 2001; Bookheimer 2002), communication between these areas tends to remain short-ranged based on the cortices’ close proximity to one another. These shorter-range, local connections may be less affected by neural underconnectivity in ASD, resulting in intact semantic processing of pictures (Sahyoun et al. 2010b). The documented overconnectivity in short-range connections (Parellada et al. 2014; Baribeau and Anagnostou 2015; Hull et al. 2017) could also explain the superior visual processing often documented in this population (Mottron et al. 2006). Lexico-semantic processing, in contrast, relies on tempo-parietal areas but is thought to recruit additional parts of the frontal cortex’s language-specific network as well (Vandenberghe et al. 1996; Cabeza and Nyberg 2000; Price 2010). This recruitment requires longer-range communication than that required for visuo-semantic processing, forming a more widely connected network for language processing than for objects. These longer-range fronto-parietal networks may be more susceptible to the consequences of underconnectivity in ASD, resulting in language difficulties (Sahyoun et al. 2010b).
EEG coherence as a measure of neural connectivity
In sum, we propose that the language-specific deficits in semantic processing in individuals with ASD (and more specifically, the findings of reduced or absent N400 effects for linguistic stimuli like words and sentences), may be a result of poor neural communication between the language and semantic systems. While this proposal fits in with prior neuroimaging findings of impaired connectivity in ASD, the temporal aspects of this relationship remain unclear. It could be that poor communication occurs in pre-semantic time windows, during the initial communication between word and semantic brain areas, in which case underconnectivity should be apparent in time windows leading up to the onset of the N400. Alternatively, the deficit could arise from poor communication at post-lexical integration steps, such as during the late positivity ERP components, in which case underconnectivity should be apparent at later time windows after the N400. Identifying the temporal focus of this underconnectivity could provide important insights into which particular aspect of neural communication is going awry during semantic processing.
While previous studies of neural connectivity in ASD have made significant gains towards understanding network communication at a broad level, they have been restricted in their ability to pinpoint the dynamic nature of connectivity. Most functional connectivity studies of language in ASD have utilized fMRI. However, because fMRI relies on changes in the blood oxygen level dependent (BOLD) signal, its temporal resolution is necessarily limited to seconds. This poor temporal resolution makes fMRI a less-than-ideal choice for questions of when neural underconnectivity occurs.
EEG has superb temporal resolution, on the order of milliseconds, and is more appropriate for questions involving the timing of neural functions. More specifically, EEG coherence is a measure of the synchronicity of oscillations, regardless of direction, between two electrodes within a designated frequency band, hypothesized to represent the communication through either long- or short-range networks throughout the brain (Fries 2005). Coherence specifically describes the extent to which two signals, within the same frequency band, share a consistent phase relationship, meaning that the waves are similarly offset from an initial starting point (Thatcher 2004; Roach and Mathalon 2008; Siegel et al. 2012; Bastos and Schoffelen 2016). Increased coherence is taken to reflect synchronized neural activation and is used as a measure of functional connectivity between brain regions. Because it maintains the precise temporal information of ERPs, EEG coherence is ideal for investigating changes in the timecourse of neural activation and connectivity. Event-related EEG coherence time-locks changes in coherence to specific events or changes in stimuli. This has been used in previous studies to highlight functional relatedness between brain areas in motor processing (Rappelsberger et al. 1994; Andrew and Pfurtscheller 1996), movement (Leocani et al. 1997), and inhibition (Shibata et al. 1998).
Here we propose to examine task- related changes in EEG coherence to pinpoint when underconnectivity occurs in the timeline of semantic processing in ASD participants. In EEG spectral analyses, different frequency bands are typically associated with different cognitive functions. The theta band (typically 3–7.5 Hz; Halgren et al. 2015; Mellem et al. 2013; Meyer et al., 2015) has been specifically associated with semantic processing and integration (Maguire and Abel 2013). Using a semantic priming paradigm with words, similar to that employed in the current study, Mellem et al. (2013) noted greater increases in anterior–posterior theta coherence for unrelated versus related words from approximately 145–425 and then 600–900 ms. They hypothesized these increases to represent the initial fronto-parietal word retrieval loop for unrelated words, and a later effect of semantic relatedness on working memory. Using a similar EEG design, Meyer et al. (2015) also found increases in EEG theta coherence during sentence reading and comprehension, specifically between left-frontal and left-parietal areas, thus contributing to the functional connection between the theta band and semantic processing. Theta power specifically (power is an estimate of the amplitude of theta oscillations) has also been found to increase when stimuli are more difficult to integrate (Klimesch et al. 1997; Brier et al. 2010), further linking theta band activity to semantic processing. Although previous studies have recorded the timecourse of semantic processing in TD populations, none have examined the differences in the context of ASD.
The current study
Here we aim to pinpoint how and why semantic integration deficits in ASD might be localized to the linguistic domain by contrasting EEG coherence during lexico-semantic and visuo-semantic processing. The timecourse of neural communication during semantic processing is of interest for this study because it could help to pinpoint the processing stage at which deficits are occurring. For example, if reduced coherence is observed in the ASD group before the onset of the N400 ERP component, it could suggest that deficits in pre-semantic communication are contributing to later semantic deficits. In contrast, if differences in coherence occur after the onset of the N400, this might suggest that differences in post-semantic integration are contributing to semantic misunderstandings. Differences in coherence at each varying timepoint could implicate the involvement of different neural connections to semantic difficulties in ASD. Both of these possibilities would lead to similar results if using fMRI to look at neural connectivity, due to its poor temporal resolution; however, EEG coherence offers the ability to tease apart these different time-locked processing stages to better characterize changes in neural connectivity.
To address this aim, the current study will reanalyze the data reported in (O’Rourke and Coderre 2021) using spectral analyses to investigate theta coherence during lexico-semantic and visuo-semantic processing. In that study we used an implicit semantic priming paradigm in which participants viewed semantically related or unrelated pairs of pictures or words (see Fig. 2). The ERP results in O’Rourke and Coderre (2021) showed an N400 effect in response to picture stimuli (i.e. more negative amplitudes for unrelated compared to related pictures) from approximately 400–500 ms in the TD group, and from approximately 200–400 ms in the ASD group. In the word condition, the TD group showed an N400 effect from approximately 400–700 ms, and the ASD group from approximately 300–600 ms.
Fig. 2 [Images not available. See PDF.]
Examples of stimuli from the experimental paradigm. Participants were shown pairs of related or unrelated words and pictures on the outlined timeline. The catch stimuli included a string of capital consonant letter (word condition) or a smiley face (picture condition)
Although both TD and ASD groups showed an N400 effect, these findings suggest that the size, timing, and duration of the N400 response could differ between them. This prior study only examined the magnitude of the N400 effect between groups and did not explore patterns of neural connectivity, which, as we have reviewed above, have been shown to be altered in ASD. Here, we will investigate time-locked changes in EEG coherence in the theta frequency band (since this frequency has been associated with semantic processing) following presentation of semantically related and unrelated words and pictures between individuals with ASD and TD individuals. We focus on the theta band specifically because of its link with semantic processing and integration in previous research (Halgren et al., 2015; Maguire and Abel, 2013; Mellem et al., 2013; Meyer et al., 2015). Although power and coherence are independent (for example, Mellem et al. (2013) find increases theta coherence but not theta power), we expect that if the brain uses theta oscillations for semantic processing then it may also use oscillations in these same frequencies to communicate semantic information.
In the TD group, we expect greater theta coherence in response to unrelated pairs of words or pictures compared to related pairs, reflecting enhanced semantic processing demands. In the ASD group, we expect smaller differences or nearly equal coherence levels between related and unrelated conditions. In concordance with previous neuroimaging literature, we expect reduced theta coherence for the ASD group particularly in left fronto-parietal connections for linguistic stimuli. Based on previous literature, we predict no differences in theta coherence across conditions or between groups during the semantic processing of pictures. Most importantly, we also explore the timecourse of changes in theta coherence to determine when differences in functional connectivity may arise. As stated above, determining whether altered connectivity occurs before or after the N400 ERP component can help pinpoint the locus of semantic processing difficulties. In our previous ERP study using this same dataset (O’Rourke and Coderre 2021), we observed N400 effects from approximately 200–500 ms in the picture condition (400–500 ms for the TD group, 200–400 ms for the ASD group) and from approximately 300–700 ms in the word condition (400–700 ms in the TD group, 300–600 ms in the ASD group). Therefore if we were to observe differences in theta coherence at very early time windows (e.g. before 200 ms for pictures or before 300 ms for words), this would suggest that differences in functional communication occur before the onset of semantic processing. In contrast, differences in very late time windows (e.g. after 500 ms for pictures or after 700 ms for words), this would suggest that post-semantic integration processes may be more affected by atypical functional connectivity in ASD.
Our paradigm included pairs of either related or unrelated pictures or words, and our hypothesis focused on the response to the second stimuli of each pair. To briefly preface the Results, we found increases in coherence that were not modulated by relatedness. We therefore also explored the responses to the first, single word or picture stimulus as opposed to limiting our analysis to the second or pairs. We observed early changes in theta coherence in response to both the first and second word in each pair, lending support for deficits in functional communication at the initial stages of word processing in ASD.
Methods
The data used here is the same as that reported in our prior study examining the N400 ERP component in this paradigm (O’Rourke and Coderre 2021). The participants and experimental methods described here can also be found in that publication.
Participants
Two groups of 20 participants each were recruited via email announcements, newspaper advertisements, and fliers at the University of Vermont and in the Burlington, VT community. Written and informed consent were obtained from each participant prior to testing. The first group consisted of 20 adults with ASD, ages 18–54 (M = 27.8, SD = 9.5), whose clinical diagnosis of autism or ASD (according to the DSM-IV or DSM-V, depending on the recency of the diagnosis/ evaluation) was confirmed using the Autism Diagnostic Observation Schedule Second Edition (ADOS-2) (Lord et al. 2012). The second group consisted of 20 typically developing (TD) adults ages 19–49 (M = 25.2, SD = 7.57). The groups did not differ on age, but the ASD group had significantly lower verbal and non-verbal IQ scores, as measured by the Kaufman Brief Intelligence Test Second Edition (KBIT-2; (Kaufman 2004). In the ASD group, a trend emerged towards lower receptive vocabulary knowledge, as measured by the Peabody Picture Vocabulary Test Fourth Edition (PPVT-4; Dunn & Dunn, 2007). Groups also differed significantly on a digit span task, used to assess working memory, such that the TD group had higher scores than the ASD group (t(36.6) = 2.99, p < 0.01). All participants, regardless of group, completed the Autism Quotient (AQ), a self-report measure of autistic traits that can be applicable to the general population (Baron-Cohen et al. 2001). As expected, the groups differed significantly on AQ scores, with the ASD group scoring higher than the TD group. Full participant demographics can be found in Table 1.
Table 1. Full participant demographics for the TD and ASD groups
TD group (n = 20) | ASD group (n = 20) | Group difference | ||
|---|---|---|---|---|
Age | 25.2 (19–49) | 27.8 (18–54) | t(36.3) = -0.96, p = 0.34 | |
PPVT | 115 (98–135) | 108 (80–132) | t(33.2) = 1.89, p = 0.07 | |
K-BIT | Verbal | 118 (101–141) | 107 (71–135) | t(33.3) = 2.54, p < 0.05 |
Non-verbal | 112 (74–132) | 101 (79–120) | t(37.5) = 2.38, p < 0.05 | |
Autism Quotient | 15.5 (5–37) | 30.1 (17–43) | t(37.5) = -6.28, p < 0.001 | |
Digit span | 12 (7–16) | 10 (4–16) | t(36.6) = 2.99, p < 0.01 | |
ADOS Module 4 | Social + communication totala | NA | 9.3 (6–16) | NA |
SA + RBB totalb | 12.3 (4–7) | |||
Calibrated severity score (CSS)c | 6.6 (3–10) | |||
The mean and range of each measure are reported. The results of independent-samples t-tests on each measure are shown in the ‘group difference’ column. Asterisks indicate statistically significant group differences (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
aThe Social + communication total score is taken from the original algorithm for Module 4 of the ADOS-2, which is included in the scoring forms
bSA = Social Affect; RBB = Restricted and Repetitive Behaviors. The SA + RBB total score is based off of a revised algorithm for Module 4 (Hus and Lord 2014), which was not included in the original release of the ADOS-2
cIn generating Calibrated Severity Scores (CSS) for Module 4, only participants ages 9–39 were included by Hus and Lord ( 2014). No calibrated scores were provided for participants older than 40, which means that technically we cannot calculate CSS for our older participants. In our sample, three participants were over 40. When considering all participants under age 40 (n = 17), the average CSS was 6.9 (range 4–10)
All participants had either normal or corrected-to-normal vision. All procedures were approved by the University of Vermont Institutional Review Board, and each participant received monetary compensation for their participation.
Task stimuli
The experimental paradigm was identical to that employed in O’Rourke and Coderre (2021). Briefly, pairs of related and unrelated words and pictures were created. Semantic relatedness was assessed using latent semantic analysis (LSA; mean LSA value for related pairs = 0.58; mean LSA value for unrelated pairs = 0.03). For more details regarding stimuli development see Coderre et al. (2017). An additional 16% of trials (16 trials per block) were “catch trials,” in which the stimulus was either a smiley face (for picture blocks) or a consonant string (for word blocks); see Fig. 2. Participants were asked to press a button on a button box whenever they saw a smiley face or consonant string. Although this procedure includes a task, this paradigm was created to assess implicit semantic integration, as participants are not asked to attend to the semantic relationship between the stimuli. Stimuli were presented in four blocks of word pairs and four blocks of picture pairs (25 related pairs, 25 unrelated pairs, 16 catch trials per block) in each of the four stimuli types (word related, word unrelated, picture related, picture unrelated), for a total of 100 trials, plus 64 catch trials in each modality. Each participant performed all four blocks of one modality followed by all four blocks of the other. The modality presented first was counterbalanced across participants.
Procedure
Stimuli were presented using E-Prime version 2.0.10.356. On each trial, a red fixation cross was presented for 400 ms, followed by the first word or picture for 1000 ms; an inter-stimulus blank screen for 300 ms; and the second word or picture for 1000 ms. Following the offset of the second stimulus a blank screen was presented for 400 ms, followed by a black fixation cross presented at an inter-trial interval ranging from 1000 to 1400 ms (mean 1200 ms).
Stimuli were presented on a Dell 21.5″ LCD monitor with a resolution of 1920 × 1080. Participants sat approximately 24″ away from the computer screen. Picture size ranged from 2.25 to 4″ in height and 2.5 to 4″ in width, yielding a visual angle between 5 and 9 degrees. Word stimuli were presented in size 28 Courier New font and ranged from 0.6 to 2.25″ in width and 0.25″ in height, yielding a visual angle between 1 and 5 degrees horizontally and 0.6 degrees vertically. During the semantic priming task, EEG data were recorded at 500 Hz using a 128-channel Geodesics Sensor net and NetStation version 5.3. Impedences were kept under 50 kΩ wherever possible. The entire experimental session lasted approximately 1.5 h including consenting, paperwork, EEG net application, and testing.
Data preprocessing
Data were preprocessed using EEGlab version 14.1.1 (Delorme & Makeig 2004) and Matlab 2017a. The data were filtered using a 0.1–50 Hz bandpass filter and re-referenced using an average reference transform. The filtered continuous data were then segmented into epochs time-locked to the onset of the first stimulus. Segments extended from 800 ms before to 3000 ms after the first stimulus (in order to capture the response to the second stimulus, presented at 1300 ms, and to provide enough data to run the time–frequency decomposition at low frequencies). Eye movement artifacts were identified and removed from the data using independent component analysis (ICA). Prior to ICA decomposition, the mean of each trial was removed (Groppe et al. 2009) and the data were reduced to 32 dimensions. After ICA decomposition, all epochs were visually identified and any epochs containing eye movements, blinks, muscle artifacts, or other noise components not removed during the ICA decomposition step were removed from the data.
Spectral analyses
EEG coherence analysis was performed using the newcrossf function in EEGlab version 14 and Matlab version 2017a (MATLAB 2010). Spectral decomposition was performed using a Morlet wavelet of 2 cycles with an expanding factor of 0.5 and a Hanning taper. Baseline correction was performed using data from 100 ms before stimulus onset. Coherence was calculated for 12 intrahemispheric electrode pairs across the scalp based on 9 electrodes taken from the 10–20 distribution system (Coben et al. 2008); see Fig. 3. These pairs were subcategorized as short-distance (left hemisphere: F3-C3, C3-P3, and P3-O1; right hemisphere: F4-C4, C4-P4, and P4-O2); medium-distance (left hemisphere: F3-P3 and C3-O1; right hemisphere: F4-P4 and C4-O2); and long-distance (left hemisphere: F3-O1; right hemisphere: F4-O2). These pairs were chosen based on a previous study of theta coherence in ASD (Coben et al. 2008) and to provide a range of electrode differences to facilitate examining short-, medium-, and long-range connections (since previous research has documented atypical connectivity in both short-range and long-range connections in ASD).
Fig. 3 [Images not available. See PDF.]
Electrode montage showing the 12 electrode pairs used for coherence analyses. Red lines indicate short-distance connections (left hemisphere: F3-C3, C3-P3, and P3-O1; right hemisphere: F4-C4, C4-P4, and P4-O2); green lines indicate medium-distance connections (left hemisphere: F3-P3 and C3-O1; right hemisphere: F4-P4 and C4-O2); and blue lines indicate long-distance connections (left hemisphere: F3-O1; right hemisphere: F4-O2)
For each pair, coherence was calculated at 393 frequencies from 2–50 Hz (approximately every 0.1 Hz) and at 300 timepoints. These parameters resulted in a time window of time–frequency decomposition from − 242 ms to 2440 ms around the first stimulus (i.e. every 8 ms). For the purposes of this study, we only examine the theta frequency band, defined as 3.5–7.5 Hz (Coben et al. 2008; Mellem et al. 2013; Halgren et al. 2015; Meyer et al. 2015), and only the first 800 ms after each stimulus presentation.
Statistical analyses
Lexico-semantic processing, as measured by the N400 ERP effect, typically occurs between approximately 300 and 500 ms (Kutas and Federmeier 2011). However, as discussed above, atypical neural communication during lexico-semantic processing in ASD may occur earlier, during the initial communication between lexical and semantic brain areas, or later in the processing stream, during post-lexical feedback processing. Because we wished to examine the full timecourse of neural connectivity underlying lexico-semantic processing in ASD, and did not wish to restrict our investigations to an a priori time window, group differences in theta coherence were evaluated by running repeated-measure ANOVAs in 100 ms windows from 100 to 800 ms after the presentation of the word or picture stimulus. This upper time limit was chosen to extend beyond the N400 to include post-semantic processing in our analyses. ANOVAs were run for each modality (picture/word) and electrode distance (short/medium/long) individually.
We first explored the effects of semantic relatedness by examining changes in theta coherence time-locked to the onset of the second stimulus (presented at 1300 ms). (Because the two stimuli in each word/picture pair were presented sequentially, relatedness cannot be established until the second stimulus is presented.) The levels of the ANOVAs in these comparisons were: group (TD/ASD), condition (related/unrelated), and hemisphere (left/right). As will be discussed, we also explored neural connectivity in response to single word/picture processing, for which we examined theta coherence time-locked to the onset of the first stimulus (presented at 0 ms). The levels of the ANOVAs in these comparisons were: group (TD/ASD) and hemisphere (left/right). (Because relatedness could not be established by only the first stimulus, condition was dropped as a factor in these analyses.) To account for group differences in demographic measures, verbal and non-verbal KBIT, PPVT, and digit span scores were included as covariates in all analyses.
Results
Response to second stimulus to evaluate effects of semantic relatedness
Word modality
Figure 4 shows the averaged theta coherence in response to the second word stimulus at each electrode pair.
Fig. 4 [Images not available. See PDF.]
Line graphs of theta coherence (averaged over all frequencies between 3 and 7.5 Hz) after the second stimulus at all intrahemispheric electrode pairs for each group and condition in the word modality. Shading represents the standard error of the mean.
Full results from the ANOVA in short-distance pairs for the word modality can be found in Table 2. From 1800 to 1900 ms (500–600 ms after presentation of the second stimulus), there was a significant interaction of group, condition, and hemisphere (F(1,38) = 4.18, p < 0.05). Follow-up ANOVAs showed a trend of an interaction of group and condition in the right hemisphere (F(1,38) = 3.21, p = 0.08) but not the left (p =− 0.81). However, upon further follow-up there were no main effects of condition in either group in the right hemisphere (all p’s > 0.13).
Table 2. F-values for the repeated-measures ANOVAs in each analysis window after the second stimulus (presented at 1300 ms) for the word modality at short-distance pairs (F3-C3, C3-P3, P3-O1, F4-C4, C4-P4, P4-O2), with a between-subjects factor of group (TD, ASD) and within-subjects factors of condition (related, unrelated) and hemisphere (left, right)
Main effect or interaction | 1400–1500 ms | 1500–1600 ms | 1600–1700 ms | 1700–1800 ms | 1800–1900 ms | 1900–2000 ms | 2000–2100 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 0.16 | 0.57 | 0.55 | < 0.01 | 0.21 | 1.33 | 1.59 |
KBIT: nonverbal | 1.57 | 1.64 | 0.22 | 0.32 | 0.06 | 0.04 | 0.12 |
PPVT | 4.58* | 0.74 | 0 | 0.50 | 2.32 | 0.89 | 0.27 |
Digit span | < 0.01 | < 0.01 | 0.11 | 0.20 | 0.94 | 0.96 | 0.72 |
Group | 0.12 | 0.14 | 0.20 | 0.72 | 0.36 | 0.14 | 0.33 |
Condition | 2.74 | 5.19* | 2.25 | 2.35 | 1.84 | 2.76 | 0.52 |
Group*Condition | 0.07 | 2.50 | 1.56 | 0.07 | 1.09 | 0.08 | 0.26 |
Hemisphere | 0.01 | 0.40 | < 0.01 | 0.15 | 0.02 | 1.29 | 2.30 |
Group*Hemisphere | 0.30 | 0.20 | 0.23 | 0.37 | 1.11 | 1.69 | 1.29 |
Condition*Hemisphere | 0.19 | 2.12 | 1.38 | 0.73 | 0.10 | 0.40 | 0.01 |
Group*Condition*Hemisphere | 0.56 | 0.19 | 1.30 | 2.62 | 4.18* | 1.43 | 0.03 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Full results from the ANOVA in medium-distance pairs for the word modality can be found in Table 3. From 1400 to 1500 ms (100–200 ms after presentation of the second stimulus) there was a significant interaction of group, condition, and hemisphere (F(1,38) = 4.55, p < 0.05). This arose from a trend towards a group and condition interaction in the left hemisphere (F(1,38) = 2.98, p = 0.09) such that the TD group showed greater theta coherence in the related condition (M = 0.05, SD = 0.11) than the unrelated condition (M = 0.02, SD = 0.12; F(1,19) = 6.75, p < 0.05), but there was no main effect of condition in the left hemisphere for the ASD group (p = 0.92). Follow-up paired-samples t-tests in the TD group at each medium-distance pair in the left hemisphere indicated a significant difference between conditions at the C3-O1 pair (t(19) = 3.17, p < 0.01) such that coherence was greater for related conditions (M = 0.04, SD = 0.10) than unrelated conditions (M = −0.01, SD = 0.09); no differences occurred at the F3-P3 pair (t(19) = 0.10, p = 0.92). Therefore, this significant effect of condition in the left hemisphere in the TD group was driven by the C3-O1 pair.
Table 3. F-values for the repeated-measures ANOVAs in each analysis window after the second stimulus (presented at 1300 ms) for the word modality at medium-distance pairs (F3-P3, C3-O1, F4-P4, C4-O2), with a between-subjects factor of group (TD, ASD) and within-subjects factors of condition (related, unrelated) and hemisphere (left, right)
Main effect or interaction | 1400–1500 ms | 1500–1600 ms | 1600–1700 ms | 1700–1800 ms | 1800–1900 ms | 1900–2000 ms | 2000–2100 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 0.33 | < 0.01 | 1.15 | 1.80 | 0.43 | 0.19 | 0.44 |
KBIT: nonverbal | 0.15 | 0.27 | 0.02 | < 0.01 | 0.66 | 1.02 | 0.19 |
PPVT | 1.71 | 0.72 | 0.65 | 0.29 | 0.08 | 0.02 | 0.04 |
Digit span | 0.24 | 0.12 | 0.42 | 2.57 | 5.30* | 6.46* | 2.27 |
Group | 0.64 | 0.89 | 0.03 | 0.39 | 0.11 | 0.22 | < 0.01 |
Condition | 5.34* | 4.25* | 1.52 | 1.6 | 8.18** | 2.32 | 0.04 |
Group*Condition | 0.01 | 0.86 | 3.29 | 2.85 | 4.13* | 1.44 | 0.69 |
Hemisphere | 1.54 | 0.43 | 0.02 | 0.03 | 1.59 | 2.50 | 1.17 |
Group*Hemisphere | 0.02 | 0.19 | 0.16 | 0.18 | 0.10 | 0.03 | 0.01 |
Condition*Hemisphere | 0.06 | 0.71 | 0.88 | 1.02 | 0.86 | 2.37 | 4.86* |
Group*Condition*Hemisphere | 4.55* | 8.87** | 5.26 | 1.82 | 2.24 | 5.94* | 2.55 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
From 1500 to 1600 ms (200–300 ms after presentation of the second stimulus) in the medium-distance pairs, there was a significant interaction of group, condition, and hemisphere (F(1,38) = 8.87, p < 0.01; Table 3) such that the ASD group had greater theta coherence in the related condition (M = 0.04, SD = 0.08) than the unrelated condition (M = 0.0004, SD = 0.09) for right-hemisphere medium-distance pairs (F(1,19) = 17.99, p < 0.001), whereas the TD group did not show a significant effect of condition in the right hemisphere (p = 0.87). Follow-up paired-samples t-tests at each medium-distance pair in the right hemisphere for the ASD group showed significant differences between related and unrelated conditions at both F4-P4 and C4-O2 pairs (all p’s < 0.05), such that the related condition had greater theta coherence than the unrelated condition.
We also noticed from visual inspection of Fig. 4 that the F3-P3 electrode pair showed a large early peak in theta coherence from approximately 100–300 ms for both conditions in the TD group but not the ASD group. Although this effect did not come out in the overall ANOVAs in this time window (i.e. there was no interaction of group and hemisphere), this may be because the F3-P3 electrode pair was only one of four medium-distance pairs being examined and the effect at this single pair was not strong enough. To follow up this observation, we ran a group by condition ANOVA at the F3-P3 cluster only from 1400 to 1600 ms (i.e. 100–300 ms after second stimulus presentation). This analysis showed a trend towards a main effect of group (F(1,34) = 3.75, p = 0.06) such that, over both conditions, the TD group showed greater theta coherence (M = 0.05, SD = 0.11) than the ASD group (M = −0.01, SD = 0.07). Interestingly, when running this same analysis just at the F4-P4 pair in the same time window, there was no main effect of group (p = 0.96) and no interaction of group and condition (p = 0.50), suggesting that the effect was specific for the left-hemisphere fronto-parietal connection.
From 1800 to 1900 ms (500–600 ms after presentation of the second stimulus) in the medium-distance pairs, there was a significant interaction of group and condition (F(1,38) = 4.13, p < 0.05; Table 3). This arose from a main effect of condition in the ASD group (F(1,19) = 10.64, p < 0.01) such that the related condition showed greater theta coherence (M = 0.01, SD = 0.08) than the unrelated condition (M = −0.03, SD = 0.08), whereas there was no main effect of condition in the TD group (p = 0.54).
From 1900 to 2000 ms (600–700 ms after presentation of the second stimulus) in the medium-distance pairs, there was a significant interaction of group, condition, and hemisphere (F(1,38) = 5.94, p < 0.05; Table 3). This arose from an interaction of group and condition in the right hemisphere (F(1,38) = 5.28, p < 0.05) such that the ASD group showed significantly greater theta coherence in the related condition (M = 0.01, SD = 0.07) than the unrelated condition (M = −0.03, SD = 0.07; F(1,19) = 7.70, p < 0.05), whereas the TD group did not show any effects of condition (p = 0.80) in the right hemisphere. Follow-up paired-samples t-tests at each medium-distance pair in the right hemisphere showed significant differences between related and unrelated conditions at both F4-P4 and C4-O2 in the ASD group (all p’s < 0.05) such that the related condition had greater theta coherence than the unrelated condition.
Full results from the ANOVA in long-distance pairs for the word modality can be found in Table 4. From 1700 to 1800 ms (400–500 ms after presentation of the second stimulus), there was a main effect of group (F(1,34) = 8.62, p < 0.01) such that, at both long-distance pairs in both hemispheres and over both conditions, the ASD group showed greater theta coherence (M = 0.02, SD = 0.08) than the TD group (M = −0.02, SD = 0.07).
Table 4. F-values for the repeated-measures ANOVAs in each analysis window after the second stimulus (presented at 1300 ms) for the word modality at long-distance pairs (F3-O1, F4-O2), with a between-subjects factor of group (TD, ASD) and within-subjects factors of condition (related, unrelated) and hemisphere (left, right)
Main effect or interaction | 1400–1500 ms | 1500–1600 ms | 1600–1700 ms | 1700–1800 ms | 1800–1900 ms | 1900–2000 ms | 2000–2100 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 4.60* | 0.63 | 0.02 | 0.07 | 0.07 | 0.19 | 1.11 |
KBIT: nonverbal | 0.16 | 0.13 | 0.28 | 0.01 | 0.06 | 0.01 | 0.98 |
PPVT | 5.33* | 5.82* | 1.38 | 0.04 | 1.21 | 0.03 | 0.01 |
Digit span | 3.52 | 4.84* | 1.43 | 0.46 | 0.91 | 1.01 | 0.96 |
Group | 0.36 | 0.06 | 3.77 | 8.62** | 4.24* | 0.17 | 0.01 |
Condition | 0.08 | 0.05 | 0.40 | 0.30 | 0.04 | 0.63 | 0.36 |
Group*Condition | 0.04 | 3.19 | 0.02 | 0.56 | 0.01 | 0.22 | 0.86 |
Hemisphere | 0.02 | 1.20 | 0.46 | 0.79 | 1.70 | 0.79 | 0.01 |
Group*Hemisphere | 0.38 | 0.03 | 0.01 | 0.14 | 0.04 | 0.06 | 0.04 |
Condition*Hemisphere | 1.73 | 3.08 | 3.31 | 2.00 | 5.35* | 7.72** | 3.47 |
Group*Condition*Hemisphere | 0.85 | 0.04 | 0.21 | 0.04 | 0.06 | < 0.01 | 1.11 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
From 1800 to 1900 ms (500–600 ms after presentation of the second stimulus) in the long-distance pairs, there was a main effect of group (F(1,34) = 4.24, p < 0.05; Table 4) such that, at both long-distance pairs in both hemispheres and over both conditions, the ASD group showed greater theta coherence (M = 0.01, SD = 0.08) than the TD group (M = −0.01, SD = 0.08).
Picture modality
Figure 5 shows the averaged theta coherence in response to the second picture stimulus at each electrode pair.
Fig. 5 [Images not available. See PDF.]
Line graphs of theta coherence (averaged over all frequencies between 3 and 7.5 Hz) after the second stimulus at all intrahemispheric electrode pairs for each group and condition in the picture modality. Shading represents the standard error of the mean
Full results from the ANOVA in short-distance pairs for the picture modality can be found in Table 5. From 1400 to 1500 ms (100–200 ms after presentation of the second stimulus), there was a main effect of group (F(1,34) = 4.63, p < 0.05) such that the TD group showed greater theta coherence (M = 0.01, SD = 0.10) than the ASD group (M = -0.01, SD = 0.09).
Table 5. F-values for the repeated-measures ANOVAs in each analysis window after the second stimulus (presented at 1300 ms) for the picture modality at short-distance pairs (F3-C3, C3-P3, P3-O1, F4-C4, C4-P4, P4-O2), with a between-subjects factor of group (TD, ASD) and within-subjects factors of condition (related, unrelated) and hemisphere (left, right)
Main effect or interaction | 1400–1500 ms | 1500–1600 ms | 1600–1700 ms | 1700–1800 ms | 1800–1900 ms | 1900–2000 ms | 2000–2100 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 0.05 | 0.67 | 1.63 | 0.69 | < 0.01 | 0.06 | 2.78 |
KBIT: nonverbal | 0.02 | 0.65 | 0.57 | 0.02 | 0.32 | 0.12 | 0.11 |
PPVT | 0.65 | 2.01 | 3.40 | 3.00 | 5.00* | 4.72* | 1.11 |
Digit span | 0.88 | 1.92 | 1.42 | 0.13 | 0.16 | 0.16 | 0.54 |
Group | 4.63* | 1.60 | 0.23 | 0.02 | 0.10 | 0.68 | 0.63 |
Condition | 0.20 | 0.27 | 1.44 | 1.75 | 3.15 | 4.97* | 5.49* |
Group*Condition | 1.16 | 0.05 | 0.02 | 0.01 | 0.04 | 0.64 | 0.14 |
Hemisphere | 0.60 | 0.16 | 0.31 | 5.00* | 4.20* | 6.89* | 5.34* |
Group*Hemisphere | 0.07 | 0.00 | 0.09 | 0.03 | 0.18 | 0.11 | 0.01 |
Condition*Hemisphere | 0.02 | 1.44 | 0.00 | 0.07 | 1.10 | 2.84 | 0.43 |
Group*Condition*Hemisphere | 0.20 | 1.74 | 2.01 | 0.03 | 5.46* | 0.38 | 0.07 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
From 1800 to 1900 ms (500–600 ms after presentation of the second stimulus) in the short-distance pairs, there was a significant interaction of group, condition, and hemisphere (F(1,38) = 5.46, p < 0.05; Table 5). This arose from an interaction of condition and hemisphere in the ASD group (F(1,19) = 6.62, p < 0.05) but not in the TD group (p = 0.40). In the ASD group, there was a main effect of condition in the right hemisphere (F(1,19) = 6.11, p < 0.05), such that the related condition showed greater theta coherence (M = −0.0002, SD = 0.08) than the unrelated condition (M = -0.02, SD = 0.07), whereas there was no effect of condition in the left hemisphere (p = 0.81).
Full results from the ANOVA in medium-distance pairs for the picture modality can be found in Table 6. From 1500 to 1600 ms (200–300 ms after presentation of the second stimulus), there was a main effect of group (F(1,34) = 4.64, p < 0.05) such that, over all electrode pairs and conditions, the TD group had greater coherence (M = 0.04, SD = 0.10) than the ASD group (M = 0.01, SD = 0.09).
Table 6. F-values for the repeated-measures ANOVAs in each analysis window after the second stimulus (presented at 1300 ms) for the picture modality at medium-distance pairs (F3-P3, C3-O1, F4-P4, C4-O2), with a between-subjects factor of group (TD, ASD) and within-subjects factors of condition (related, unrelated) and hemisphere (left, right)
Main effect or interaction | 1400–1500 ms | 1500–1600 ms | 1600–1700 ms | 1700–1800 ms | 1800–1900 ms | 1900–2000 ms | 2000–2100 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 0.34 | 0.56 | 1.53 | 0.40 | 0.83 | 3.01 | 5.85* |
KBIT: nonverbal | < 0.01 | 0.05 | 0.02 | 0.05 | 2.03 | 1.15 | 0.32 |
PPVT | < 0.01 | 0.07 | 0.00 | 0.01 | 0.01 | 1.79 | 2.31 |
Digit span | 4.63* | 3.15 | 1.62 | 0.18 | 0.30 | 0.23 | 0.63 |
Group | 1.67 | 4.64* | 2.64 | 0.99 | 0.77 | 0.82 | 0.04 |
Condition | 0.57 | 0.80 | 0.81 | 0.99 | 3.37 | 0.45 | 1.16 |
Group*Condition | 1.54 | 1.17 | 2.97 | 2.89 | 2.60 | 0.80 | < 0.01 |
Hemisphere | 1.95 | 1.63 | 3.51 | 4.87* | 2.05 | 1.34 | 1.53 |
Group*Hemisphere | 0.35 | 0.63 | 1.86 | 1.92 | 2.07 | 1.78 | 0.48 |
Condition*Hemisphere | 0.10 | 0.20 | 0.06 | 0.05 | < 0.01 | 0.65 | 0.39 |
Group*Condition*Hemisphere | 1.60 | 0.89 | 1.51 | 0.54 | 0.05 | 1.63 | 1.19 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Full results from the ANOVA in long-distance pairs for the picture modality can be found in Table 7. There were no significant main effects of or interactions with group in any time window.
Table 7. F-values for the repeated-measures ANOVAs in each analysis window after the second stimulus (presented at 1300 ms) for the picture modality at long-distance pairs (F3-O1, F4-O2), with a between-subjects factor of group (TD, ASD) and within-subjects factors of condition (related, unrelated) and hemisphere (left, right)
Main effect or interaction | 1400–1500 ms | 1500–1600 ms | 1600–1700 ms | 1700–1800 ms | 1800–1900 ms | 1900–2000 ms | 2000–2100 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 0.18 | 0.48 | 0.27 | 0.55 | 0.02 | 0.50 | 0.13 |
KBIT: nonverbal | 0.47 | 0.56 | 0.88 | 0.29 | < 0.01 | 0.01 | 0.32 |
PPVT | 0.78 | 0.20 | 0.59 | 0.98 | 0.61 | 2.82 | 4.57* |
Digit span | 4.97* | 2.26 | 1.93 | 0.74 | 0.13 | 0.15 | 0.03 |
Group | 0.27 | 0.93 | 0.21 | 0.05 | 0.05 | 0.06 | 0.33 |
Condition | 1.89 | 0.86 | 0.28 | 0.04 | 0.18 | 0.11 | 0.09 |
Group*Condition | 0.03 | 0.24 | 1.27 | 0.63 | 0.21 | 0.83 | 0.85 |
Hemisphere | 0.06 | 0.92 | 0.87 | 0.02 | 0.29 | 1.05 | 0.63 |
Group*Hemisphere | < 0.01 | 0.01 | 0.19 | 0.02 | 0.28 | 0.74 | 0.18 |
Condition*Hemisphere | 0.70 | 0.36 | 0.56 | 2.04 | 1.22 | 0.84 | 0.92 |
Group*Condition*Hemisphere | 0.27 | 0.12 | 0.49 | 0.06 | 1.92 | 0.04 | 2.76 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Summary: response to second stimulus
To summarize, we first examined changes in theta coherence time-locked to the onset of the second stimulus to investigate effects of semantic relatedness.
In the word modality, there were no significant group differences in short distance pairs. In medium-distance pairs, the TD group showed an early effect of condition (related > unrelated) from 100 to 200 ms at C3-O1 (left hemisphere medium-distance pair) whereas the ASD group did not. At the next time window from 200 to 300 ms, this effect was virtually flipped, with the ASD group showing effects of condition (related > unrelated) in both right-hemisphere medium-distance pairs, whereas the TD group did not. At later time windows, the ASD group showed effects of condition (related > unrelated) over all medium-distance pairs in both hemispheres from 500 to 600 ms, and in the right hemisphere from 600 to 700 ms, whereas the TD group did not show any effects of condition in these time windows. In long-distance pairs, the ASD group showed greater coherence than the TD group, when collapsed over condition, across both hemispheres from 400 to 600 ms. Overall, for the word condition the results were fairly scattered. Although there were some early effects of condition (100–300 ms), these were not consistent between groups and/or hemispheres.
Perhaps the most interesting finding from the word modality is that we noticed from visual inspection of Fig. 4 that the F3-P3 medium distance pair showed large early peaks in theta coherence for the TD group but not for the ASD group. When running follow-up ANOVAs just at the F3-P3 pair, there was a trend of greater overall coherence in the TD group than the ASD group from 100 to 300 ms, whereas this group difference did not occur for the analogous right-hemisphere pair (F4-P4). As this left-hemisphere fronto-parietal connection is critical for language, this increase in theta coherence could be reflecting pre-semantic communication that is atypical in the ASD group.
In the picture modality, the TD group showed greater coherence than the ASD group from 100 to 200 ms over all short-distance pairs in both hemispheres. From inspection of Fig. 5, this appeared to be driven mostly by the P3-O1 and P4-O2 pairs. From 500 to 600 ms there was an effect of condition (related > unrelated) in right hemisphere short-distance pairs for the ASD group but not for the TD group. In medium-distance pairs the TD group showed greater coherence than ASD over both conditions over all medium-distance pairs in both hemispheres from 200 to 300 ms. There were no group differences in long-distance pairs. In sum, for the picture condition there were some early effects from 100 to 300 ms, with more coherence for the TD group than the ASD group (but no effects of condition) in both short and medium distance pairs in both hemispheres. There were also some effects of condition at later time windows from 500 to 600 ms for the ASD group (related > unrelated) but not the TD group.
Response to first stimulus to evaluate initial early peak
As mentioned above, one interesting finding from the results in response to the second stimulus was the initial peak in theta coherence at left fronto-parietal connections (F3-P3) in the TD group but not the ASD group. As this effect was consistent across conditions (i.e. there were no differences in theta coherence between related and unrelated conditions), one possibility is that this initial increase in coherence reflects early, pre-semantic communication of linguistic information along left-lateralized fronto-parietal pathways. If this is the case, then a similar increase in coherence should occur in response to any single word, and therefore it should also be observable in response to the first stimulus as well. We had initially run our analyses in response to the second stimulus to investigate effects of relatedness. However, because the two stimuli in each word/picture pair were presented sequentially in the current paradigm (see Fig. 2), we can also evaluate theta coherence in response to the first stimulus, before relatedness has been established. We therefore reran our analyses time-locked to the onset of the first stimulus, averaged over related and unrelated conditions.
Word modality
Figure 6 shows the averaged theta coherence in response to the first word stimulus at each electrode pair.
Fig. 6 [Images not available. See PDF.]
Line graphs of theta coherence (averaged over all frequencies between 3 and 7.5 Hz) after the first stimulus at all intrahemispheric electrode pairs for each group in the word modality, averaged over related/unrelated conditions. Shading represents the standard error of the mean
Full results from the ANOVA in short-distance pairs for the word modality can be found in Table 8. From 200 to 300 ms there was a significant main effect of group (F(1,34) = 5.89, p < 0.05). Across all electrode pairs, the TD group showed greater theta coherence (M =− 0.01, SD = 0.09) than the ASD group (M = − 0.01, SD = 0.08).
Table 8. F-values for the repeated-measures ANOVAs in each analysis window after the first stimulus for the word modality at short-distance pairs (F3-C3, C3-P3, P3-O1, F4-C4, C4-P4, P4-O2), with a between-subjects factor of group (TD, ASD) and a within-subjects factor of hemisphere (left, right)
Main effect or interaction | 100–200 ms | 200–300 ms | 300–400 ms | 400–500 ms | 500–600 ms | 600–700 ms | 700–800 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 0.30 | 0.64 | 0.22 | 0.34 | 0.12 | 0.21 | 0.72 |
KBIT: nonverbal | 1.51 | 1.28 | 0.77 | 0.76 | 0.18 | 0.08 | 0.21 |
PPVT | 3.38 | 0.78 | < 0.01 | 0.03 | 0.02 | 0.41 | 0.90 |
Digit span | 0.22 | 0.01 | 0.30 | 0.14 | 0.03 | 0.02 | 0.71 |
Group | 3.26 | 5.89* | 4.12 | 0.92 | 0.39 | 0.95 | 1.47 |
Hemisphere | 0.15 | 0.26 | 0.63 | 1.85 | < 0.01 | 0.54 | 2.09 |
Group*Hemisphere | 0.65 | 0.14 | 0.03 | 0.25 | 0.52 | 1.61 | 0.27 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Full results from the ANOVA in medium-distance pairs for the word modality can be found in Table 9. There were no main effects of or interactions with group. However, inspection of the waveforms in response to the first word stimulus (Fig. 6) again showed an early peak in theta coherence (approximately 100–300 ms) at the F3-P3 connection in the left hemisphere for the TD group but not the ASD group. This pattern is similar to that observed in response to the second word stimulus (see Fig. 4). Therefore we once again followed up this observation by running an independent-samples t-test at the F3-P3 cluster only from 100 to 300 ms after first stimulus presentation. This analysis showed a trend of a group difference (t(35.2) = 1.90, p = 0.07), such that the TD group had greater coherence (M = 0.05, SD = 0.10) than the ASD group (M = − 0.001, SD = 0.07). As in the prior comparison, there were no group differences at the analogous right-hemisphere medium distance pair, F4-P4 (p = 0.86), once again suggesting that the effect was specific for the left-hemisphere fronto-parietal connection.
Table 9. F-values for the repeated-measures ANOVAs in each analysis window after the first stimulus for the word modality at medium-distance pairs (F3-P3, C3-O1, F4-P4, C4-O2), with a between-subjects factor of group (TD, ASD) and a within-subjects factor of hemisphere (left, right)
Main effect or interaction | 100–200 ms | 200–300 ms | 300–400 ms | 400–500 ms | 500–600 ms | 600–700 ms | 700–800 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 1.25 | 0.03 | 0.12 | 0.09 | 0.10 | 0.01 | 0.10 |
KBIT: nonverbal | < 0.01 | 0.25 | 0.36 | 0.77 | 1.85 | 2. 94 | 1.47 |
PPVT | 3.83 | 0.21 | 0.32 | 0.45 | 0.12 | 0.29 | 0.03 |
Digit span | 0.61 | 0.49 | 0.02 | 0.06 | 0.32 | 0.28 | 0.21 |
Group | 0.05 | 0.07 | 0.16 | 0.01 | 0.04 | 0.15 | 0.02 |
Hemisphere | 0.85 | 0.02 | 0.02 | 0.37 | 1.10 | 0.04 | 0.05 |
Group*Hemisphere | 0.78 | 0.41 | 0.29 | 0.96 | 1.20 | 0.85 | 0.94 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Full results from the ANOVA in long-distance pairs for the word modality can be found in Table 10. From 300 to 400 ms there was a significant main effect of group (F(1, 34) = 5.97, p < 0.05), such that across all electrode pairs and conditions, the ASD group showed greater theta coherence (M = 0.03, SD = 0.08) than the TD group (M = − 0.004, SD = 0.05).
Table 10. F-values for the repeated-measures ANOVAs in each analysis window after the first stimulus for the word modality at long-distance pairs (F3-O1, F4-O2), with a between-subjects factor of group (TD, ASD) and a within-subjects factor of hemisphere (left, right)
Main effect or interaction | 100–200 ms | 200–300 ms | 300–400 ms | 400–500 ms | 500–600 ms | 600–700 ms | 700–800 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 5.50* | 2.68 | < 0.01 | < 0.01 | 0.04 | 0.04 | 0.31 |
KBIT: nonverbal | 0.02 | 0.86 | 0.13 | 1.19 | 0.12 | 0.01 | 0.33 |
PPVT | 11.00** | 10.40** | 1.68 | 1.83 | 1.91 | 0.04 | 0.01 |
Digit span | 0.43 | 0.57 | 1.34 | 3.33 | 11.48** | 8.87** | 2.32 |
Group | 0.25 | 1.62 | 5.97* | 2.88 | 0.59 | 0.99 | 2.58 |
Hemisphere | 1.43 | 1.37 | 0.62 | 0.17 | 0.81 | 0.32 | 0.38 |
Group*Hemisphere | 0.18 | < 0.01 | 0.71 | 0.22 | 0.18 | 0.45 | 0.77 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
From 800 to 900 ms there was a significant main effect of group in the long-distance pairs (F(1, 34) = 6.69, p < 0.05; Table 10). Across all electrodes and conditions, the ASD group showed greater theta coherence (M = 0.003, SD = 0.06) than the TD group (M = -0.015, SD = 0.05).
From 900 to 1000 ms there was a significant main effect of group in the long-distance pairs (F(1, 34) = 6.65, p < 0.05; Table 10). Across all electrode pairs and conditions, the ASD group showed more theta coherence (M = − 0.006, SD = 0.05) than the TD group (M = − 0.02, SD = 0.05).
Picture modality
Figure 7 shows the averaged theta coherence in response to the first word stimulus at each electrode pair.
Fig. 7 [Images not available. See PDF.]
Line graphs of theta coherence (averaged over all frequencies between 3 and 7.5 Hz) after the first stimulus at all intrahemispheric electrode pairs for each group in the picture modality, averaged over related/unrelated conditions. Shading represents the standard error of the mean
Full results from the ANOVA in short-distance pairs for the picture modality can be found in Table 11. There were no significant main effects of or interactions with group in any time window.
Table 11. F-values for the repeated-measures ANOVAs in each analysis window after the first stimulus for the picture modality at short-distance pairs (F3-C3, C3-P3, P3-O1, F4-C4, C4-P4, P4-O2), with a between-subjects factor of group (TD, ASD) and a within-subjects factor of hemisphere (left, right)
Main effect or interaction | 100–200 ms | 200–300 ms | 300–400 ms | 400–500 ms | 500–600 ms | 600–700 ms | 700–800 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 0.09 | 0.05 | 0.41 | 0.05 | < 0.01 | 0.29 | 0.35 |
KBIT: nonverbal | 0.86 | 0.40 | 1.07 | 0.78 | 2.09 | 1.70 | 1.30 |
PPVT | 0.87 | 1.57 | 0.36 | 0.43 | 0.95 | 1.69 | 1.08 |
Digit span | < 0.01 | 0.31 | 0.69 | 1.27 | 1.39 | 0.45 | 0.67 |
Group | 1.39 | 0.69 | 0.31 | 0.01 | 0.01 | 0.51 | 0.39 |
Hemisphere | 0.01 | 0.88 | 3.34 | 1.60 | 1.00 | 1.26 | 0.97 |
Group*Hemisphere | 0.27 | 0.05 | 0.30 | < 0.01 | 0.01 | 0.55 | 0.37 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Full results from the ANOVA in medium-distance pairs for the picture modality can be found in Table 12. From 700 to 800 ms there was a significant interaction of group and hemisphere (F(1, 38) = 4.92, p < 0.05). However, upon follow-up testing there were no significant effects of group in either hemisphere, nor at any individual medium-distance electrode pairs.
Table 12. F-values for the repeated-measures ANOVAs in each analysis window after the first stimulus for the picture modality at medium-distance pairs (F3-P3, C3-O1, F4-P4, C4-O2), with a between-subjects factor of group (TD, ASD) and a within-subjects factor of hemisphere (left, right)
Main effect or interaction | 100–200 ms | 200–300 ms | 300–400 ms | 400–500 ms | 500–600 ms | 600–700 ms | 700–800 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 3.70 | 3.42 | 1.79 | 0.97 | 0.54 | 0.32 | 0.24 |
KBIT: nonverbal | 0.02 | 0.20 | 1.23 | 0.26 | 0.16 | 0.13 | 0.41 |
PPVT | 0.19 | 0.11 | 0.82 | 0.56 | 0.10 | 0.02 | 0.02 |
Digit span | 2.73 | 0.47 | 0.05 | 0.99 | 0.83 | 2.42 | 1.28 |
Group | 0.12 | 0.62 | 0.19 | 0.10 | 0.35 | < 0.01 | 0.03 |
Hemisphere | 0.10 | 0.39 | 0.15 | 0.25 | 0.28 | 0.43 | 0.89 |
Group*Hemisphere | 0.13 | 0.13 | 0.55 | 0.59 | 0.31 | 1.94 | 4.92* |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Full results from the ANOVA in long-distance pairs for the picture modality can be found in Table 13. There were no significant main effects of or interactions with group in any time window.
Table 13. F-values for the repeated-measures ANOVAs in each analysis window after the first stimulus for the picture modality at long-distance pairs (F3-O1, F4-O2), with a between-subjects factor of group (TD, ASD) and a within-subjects factor of hemisphere (left, right)
Main effect or interaction | 100–200 ms | 200–300 ms | 300–400 ms | 400–500 ms | 500–600 ms | 600–700 ms | 700–800 ms |
|---|---|---|---|---|---|---|---|
KBIT: verbal | 1.44 | 0.81 | 1.28 | 0.09 | 0.07 | 0.03 | 0.54 |
KBIT: nonverbal | 1.45 | 1.40 | 0.79 | 2.01 | 2.17 | 1.15 | 1.05 |
PPVT | 0.49 | 0.93 | 1.32 | 1.87 | 1.88 | 3.78 | 1.11 |
Digit span | 1.72 | 0.95 | 0.10 | 0.24 | 0.77 | 3.26 | 2.85 |
Group | 0.12 | 0.49 | 0.06 | 0.31 | 0.31 | 0.08 | 0.02 |
Hemisphere | < 0.01 | 0.03 | 0.43 | 2.13 | 2.08 | 0.88 | 0.31 |
Group*Hemisphere | 0.14 | 0.73 | 1.54 | 1.21 | 0.59 | 1.17 | 3.35 |
Verbal and non-verbal K-BIT, PPVT, and digit span scores were included as covariates. Main effects of or interactions with group are highlighted in bold. Asterisks indicate statistically significant results (* = p < 0.05; ** = p < 0.01; *** = p < 0.001)
Summary: response to first stimulus
To summarize, we also examined changes in theta coherence time-locked to the onset of the first stimulus to investigate effects of semantic processing of single words and pictures.
In the word modality, the TD group showed greater theta coherence in early time windows (200–300 ms) than ASD across all short-distance electrode pairs in both hemispheres. For long distance pairs in words, the ASD group showed more coherence than the TD group in both left and right hemispheres across both early (300–400 ms) and late (800–1000 ms) time windows. Visual inspection of Fig. 6 suggests that this finding is driven mostly by F4-O2, which shows more sustained theta coherence in the ASD group following the initial early peak, as well as at later time windows.
Although there were no significant main effects of group in the overall analysis for medium-distance electrode pairs, visual inspection of Fig. 6 once again indicated large early peaks in theta coherence at the F3-P3 pair. Follow-up analyses at this pair showed a trend towards TD having greater coherence than ASD, whereas there were no group differences at F4-P4. This pattern is similar to that observed in response to the second stimulus.
In the picture modality, there were no significant group effects for any electrode pairs across short, medium, or long distances.
Discussion
In this study, we compared EEG theta coherence patterns between visuo-semantic and lexico-semantic processing, and between TD and ASD individuals. We expected to find decreased fronto-parietal coherence in the ASD group for linguistic stimuli, but no differences between TD and ASD coherence during the semantic processing of pictures.
Effects of relatedness
We had predicted that theta coherence would be greater for unrelated pairs of pictures and words compared to related pairs in TD individuals. This prediction was made based on prior reports that theta power is increased when stimuli are more difficult to integrate (Klimesch et al. 1997; Brier et al. 2010), as should be the case with unrelated stimuli, and from prior studies noting greater theta coherence for unrelated stimuli compared to related stimuli in a semantic priming task (Mellem et al. 2013). However, our data actually showed the opposite pattern: theta coherence was often greater for the related conditions compared to the unrelated conditions.
Whereas we had expected to find increases in coherence in response to unrelated stimuli as the brain recruits additional areas and tries to compensate for a lack of semantic fluency, it is possible that this increase occurs at a later time window than we expected, and that our time windows only capture the initial lapse. Based on the fact that we see such early changes in theta coherence, an additional possibility is that coherence in this frequency band may not represent semantic integration or the evaluation of semantic relatedness (since we saw increases in theta coherence in response to the first word in the stimulus pair, before relatedness had been established) but rather early communication between brain areas during the initial stages of semantic processing, such as the binding of tokens to types (see below). While theta has been linked to semantic processing in previous studies, perhaps the specific function of semantic integration is better captured by oscillations in other frequency bands, or by other time–frequency metrics not investigated here. It is also possible that the ASD group utilizes a different pathway for processing related and unrelated stimuli that theta coherence cannot encompass. While our interpretations here remain speculative, in any case these findings indicate that EEG coherence can provide information beyond that captured by EEG power or ERP amplitudes; future studies should expand upon these findings to further characterize patterns of event-related changes in theta coherence during semantic integration.
Group differences in words and pictures
Acknowledging our finding that theta coherence was actually enhanced for related pairs than for unrelated pairs, overall there were no clear patterns of group differences in theta coherence with regards to relatedness. For instance, at medium distance pairs in the word modality, from 1400 to 1500 ms (100–200 ms after presentation of the second stimulus), we found an effect of condition (related > unrelated) in the TD group at the left hemisphere C3-O1 pair. At the next time window (200–300 ms after presentation of the second stimulus), the direction of the group effect was reversed, with the ASD group showing effects of condition (related > unrelated) across medium distance pairs in the right hemisphere (F4-P4 and C4-O2), whereas the TD group showed no effects.
These findings do not support our hypothesis that effects would be centered around the left fronto-parietal F3-P3 pair: significant differences were often lateralized and opposite across time windows, hemispheres, groups, and pairs. This could suggest that language may be less left lateralized in ASD than in TD, as has been found by prior studies (Fletcher et al. 2010; Wan et al. 2012; Nielsen et al. 2014; Moseley et al. 2016), or that the processing loop each group uses to interpret semantic stimuli may differ. The timing of activity increases is also different between ASD and TD groups. For example, across medium-distance pairs in words after the second stimulus, the group showing significant effects of condition flipped between 1400–1500 ms and 1500–1600 ms: from 1400 to 1500 ms, the TD group showed a significant effect while the ASD did not, but from 1500 to 1600 ms, the ASD group showed a significant effect while the TD group did not. Although both groups showed the same effect and direction (related > unrelated), this occurred in different time windows, which could signify that the time course of semantic processing in ASD is fundamentally different than in TD. Latencies and discrepancies of the N400 in ASD have been found in other studies (Verbaten et al. 1991; Strandburg et al. 1993; Dunn et al. 1999; Valdizán et al. 2003; Dunn and Bates 2005; Braeutigam et al. 2008; McCleery et al. 2010; Pijnacker et al. 2010; Coderre et al. 2017), but the timecourse of semantic processing has yet to be studied in the specific context of theta coherence.
Contrary to our hypothesis, we did find significant group effects in the picture modality. At short-distance pairs in the right hemisphere from 1800 to 1900 ms (500–600 ms after presentation of the second stimulus), related pairs of pictures elicited greater theta coherence than unrelated pairs in the ASD group alone. It is possible that these findings show not that the ASD group had greater coherence levels than the TD group, but that within the ASD group, there is a significant difference between coherence levels when processing related stimuli vs. unrelated stimuli, while the TD group maintains more stable levels of coherence regardless of stimuli relatedness. Further research is needed to confirm this proposal.
Early peaks in theta coherence
Although we did not observe the expected group differences in theta coherence changes in response to relatedness, our most interesting finding is the visual observation of group differences in early (100–300 ms) peaks in theta coherence at the F3-P3 electrode pair, which represents a left- hemisphere fronto-parietal connection. As can be seen in both Figs. 4 and 6, the TD group showed initial peaks in theta coherence in response to both the first and second word stimulus across many electrode pairs. However, at the F3-P3 pair in the left hemisphere, the ASD group did not show this same peak in theta coherence. This group difference was absent in the right hemisphere—there were no group differences in early theta coherence from F4-P4, for either first or second stimuli—and was restricted to the word modality—there were no group differences in early theta coherence for picture stimuli. Follow-up statistical analyses confirmed that, although the effects were only statistical trends, the ASD group showed reduced theta coherence in this left-hemisphere fronto-parietal connection in response to linguistic stimuli compared to the ASD group. The fact that this trend towards reduced connectivity in the ASD group occurred at the F3-P3 pair only (and not in the analogous right-hemisphere electrode pair) is in line with previous research documenting underconnectivity in left fronto-parietal regions (Just 2004; Kana et al. 2006; Coben et al. 2008; Schipul et al. 2011). Furthermore, the fact that we observed this potential group difference only in the linguistic modality supports our prediction that neural connectivity during lexico-semantic processing is selectively impaired, as the same underconnectivity is not observed during visuo-semantic processing. Our findings thus confirm previous data documenting underconnectivity in left fronto-parietal connections during language processing in individuals with ASD.
Our study specifically sought to identify the temporal locus of underconnectivity using EEG coherence analysis, since pinpointing when disruptions in connectivity occur may provide important insights into what component processes are being affected. Our data identified differences in a very early time window: 100–300 ms after stimulus presentation, clearly before the onset of the N400 ERP component. This suggests that it is early, pre-semantic neural communication that is affected during linguistic processing in ASD.
To frame this finding in the broader neurolinguistic literature, we turn to a recent model proposed by Baggio (2018). Baggio puts forth an elaborative and comprehensive model of semantic processing in the brain in which he proposes the existence of an “R-system” of relational semantics, which primarily generates the N400 ERP component. Linking together electrophysiology data, neuroimaging data, and theoretical semantics, he proposes that information first flows from sensory input regions in the visual and/or temporal cortices to two specific regions of the temporal cortex: the posterior middle superior temporal gyrus (pMSTG), which houses lexical interface codes that mediate access to and retrieval of distributed lexical knowledge, and the anterior temporal lobe (ATL), which houses amodal semantic concepts. This initial flow of information happens within approximately 250 ms. These regions are involved in activating lexical (semantic) types, which are abstract and general semantic concepts.
From these temporal regions, information then flows forward to the left inferior frontal gyrus (LIFG), enabling bottom-up binding processes and reflecting the ascending phase of the N400 ERP component. The LIFG takes the representation of lexical types and binds it to a token representation, which is a more specific instance of a concept. Information reaches the LIFG at approximately 400 ms, representing the peak of the N400 ERP component. From the LIFG, information then flows back towards the pMSTG to enable top-down binding and strengthen the relations between the word and its context, generating a type-based context representation in the temporal lobe. This backwards flow of information is reflected as the descending phase of the N400 component.
In our data, we observed surprisingly early increases in theta coherence in the TD group, in response to both word stimuli, from approximately 100–300 ms, which had returned to baseline by 400 ms (see Figs. 4 and 6). This suggests that these peaks in theta coherence may be reflecting the initial temporal-to-frontal flow of information as currents are sent from the sensory cortices and orthographic input regions to the pMSTG and ATL, and then forward to the LIFG. According to Baggio’s model, functionally, this can be pinpointed as the access of lexical types and the communication of information forward to generate token representations. However, the fact that we did not see sustained increases in theta coherence beyond approximately 400 ms, when information would have been flowing back from the LIFG to temporal cortices, suggests that this theta coherence is representing early semantic activation, possibly prior to the binding process and the representations of tokens that the LIFG is responsible for. Of course, the poor spatial resolution of EEG makes this a tentative interpretation of our data; more evidence is needed that brings in other techniques such as fMRI or MEG with superior spatial resolution (see next section). However, the timing of the visually observed differences indicates that early neural communications in response to word processing might be atypical in individuals with ASD.
Even more intriguing is that we did not observe this same early increase in theta coherence in left-hemisphere fronto-parietal connections in the ASD group. Interpreting this finding in light of Baggio’s model once again would suggest that the difficulties with semantic processing in individuals with ASD may stem from early binding processes, such as during the conversion from lexical types to tokens. Moreover, poor neural communication or less efficient communication between temporal and frontal brain areas during this early time window may account for the findings of previous studies that have reported reduced or absent N400 effects in individuals with ASD: if information is not effectively sent to the LIFG for binding and token representation, this would result in reduced or absent N400 effects in the ERP signature, as has indeed been observed in many studies (Verbaten et al. 1991; Strandburg et al. 1993; Dunn et al. 1999; Valdizán et al. 2003; Dunn and Bates 2005; Braeutigam et al. 2008; McCleery et al. 2010; Pijnacker et al. 2010; Coderre et al. 2017).
Future directions
Our data has demonstrated the utility of EEG and its superb temporal resolution in identifying a potential temporal locus of underconnectivity during lexico-semantic processing in individuals with ASD. Although we observed the same left-lateralized fronto-parietal underconnectivity in the ASD group in response to both the first and second word stimulus, we acknowledge that group differences for both of these analyses were statistical trends. More research is needed to replicate these findings using methods with precise temporal resolution, such as EEG or MEG. In addition, our conclusions regarding the precise neural connections affected are speculative given the poor spatial resolution of EEG. Future research using methods with higher spatial resolution, such as fMRI or MEG, is needed to further investigate the areas involved with these findings. For example, extending the implicit semantic priming paradigm used in this study to fMRI would be necessary to confirm that individuals with ASD experience reduced connectivity in left fronto-parietal connections during semantic processing of words. Although this result has been found before using other paradigms (Just 2004; Kana et al. 2006; Coben et al. 2008; Schipul et al. 2011), no other study has used a paradigm similar to that employed here. fMRI data, and particularly functional connectivity data, collected during this paradigm would provide complementary evidence to help localize the spatial extent of underconnectivity. It is also possible that individuals with ASD rely on completely different semantic processing loops than TD individuals, which could account for the difficulties individuals with ASD often face with integrating semantic stimuli into a comprehensive understanding (Tager‐Flusberg 1991; Verbaten et al. 1991; Valdizán et al. 2003; Kamio et al. 2007; Lau et al. 2008; Pijnacker et al. 2010). Functional connectivity data from fMRI would help identify whether and how these loops differ during semantic processing tasks.
Extension to MRI also offers the possibility of utilizing structural connectivity data to confirm our findings. Of particular note are our findings in the F3-P3 pair, thought to reflect a similar neural connection as the arcuate fasciculus. Due to its fronto-parietal location and known implications in language and ASD (Ben Bashat et al. 2007; Catani and Mesulam 2008; Lebel and Beaulieu 2009; Fletcher et al. 2010; Yeatman et al. 2011; Wan et al. 2012; Lopez-Barroso et al. 2013; Joseph et al. 2014; Roberts et al. 2014; Moseley et al. 2016), the arcuate fasciculus is an excellent candidate for such follow-up MRI research. Structural properties of the arcuate fasciculus have been studied in the context of ASD (Glasser and Rilling 2008; Fletcher et al. 2010; Yeatman et al. 2011; Wan et al. 2012; Joseph et al. 2014; Roberts et al. 2014; Moseley et al. 2016), but future MRI studies could investigate additional, more minute structural connections of the tract, thus contributing to the underconnectivity hypotheses. It would be especially beneficial for this research to be longitudinal, considering that ASD is a developmental disorder whose expression can change over time. Studying the structural and functional evolution of the arcuate fasciculus could help pinpoint the root of these processing differences and could even help target intervention strategies. It is unknown if a critical development period exists for the arcuate fasciculus, but if one was discovered, it could drastically improve the way we help those with ASD learn language and interact with the world.
Conclusions
This study analyzed EEG coherence data in the theta frequency band during a semantic priming paradigm to examine neural connectivity patterns during lexico-semantic and visuo-semantic processing in individuals with and without ASD. Our main finding of interest was a consistent pattern of underconnectivity in the ASD group compared to the TD group, in line with previous findings. More specifically, this underconnectivity was restricted to left-hemisphere fronto-parietal connections and only appeared in response to words; no noteworthy group differences occurred in right-hemisphere fronto-parietal connections or in response to pictures. Furthermore, this underconnectivity occurred at very early time windows, 100–300 ms after presentation of word stimuli, suggesting deficits in neural communication during pre-semantic linguistic processes.
Authors' contributions
AC analyzed and interpreted the data; and drafted the manuscript. EC conceptualized and designed the study; obtained funding; collected, analyzed and interpreted the data; and drafted the manuscript. All authors read and approved the final manuscript.
Funding
This research was supported by the College of Nursing and Health Sciences Research Incentive Grant at the University of Vermont.
Availability of data and material
Data will be made available to interested researchers upon request.
Code availability
Code will be made available to interested researchers upon request.
Declarations
Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Ethical approval
All procedures were approved by the University of Vermont Institutional Review Board.
Consent to participate
Written informed consent was obtained from all participants before testing.
Consent for publication
All authors have agreed with the content and have given explicit consent to publish this manuscript.
In keeping with prior literature, we define “linguistic” stimuli as involving language, whether written or spoken, whereas “non-linguistic” stimuli do not involve language. Here, we will use pictures as the “non-linguistic” or “visual” stimuli and written words as the “linguistic” stimuli (even though written language is also visual).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
Anderson, JS; Druzgal, TJ; Froehlich, A et al. Decreased interhemispheric functional connectivity in autism. Cereb Cortex; 2011; 21, pp. 1134-1146. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20943668][DOI: https://dx.doi.org/10.1093/cercor/bhq190]
Andrew, C; Pfurtscheller, G. Event-related coherence as a tool for studying dynamic interaction of brain regions. Electroencephalogr Clin Neurophysiol; 1996; 98, pp. 144-148.[COI: 1:STN:280:DyaK287ntlCgug%3D%3D] [DOI: https://dx.doi.org/10.1016/0013-4694(95)00228-6] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/8598174]
Baggio, G. Meaning in the Brain; 2018; MIT Press: [DOI: https://dx.doi.org/10.7551/mitpress/11265.001.0001]
Baribeau, D; Anagnostou, E. Fatemi, SH. Neuroimaging in autism spectrum disorders. The molecular basis of autism; 2015; Springer: pp. 117-152. [DOI: https://dx.doi.org/10.1007/978-1-4939-2190-4_7]
Baron-Cohen S, Wheelwright S, Skinner R, et al (2001) The Autism-Spectrum Quotient (AQ): Evidence from Asperger Syndrome/High-Functioning Autism, Males and Females, Scientists and Mathematicians. J Autism Dev Disord 31: 5–17
Barttfeld, P; Wicker, B; Cukier, S et al. A big-world network in ASD: dynamical connectivity analysis reflects a deficit in long-range connections and an excess of short-range connections. Neuropsychologia; 2011; 49, pp. 254-263. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21110988][DOI: https://dx.doi.org/10.1016/j.neuropsychologia.2010.11.024]
Bastos, AM; Schoffelen, J-M. A tutorial review of functional connectivity analysis methods and their interpretational pitfalls. Front Syst Neurosci; 2016; 9, pp. 1-23. [DOI: https://dx.doi.org/10.3389/fnsys.2015.00175]
Ben Bashat, D; Kronfeld-Duenias, V; Zachor, DA et al. Accelerated maturation of white matter in young children with autism: a high b value DWI study. Neuroimage; 2007; 37, pp. 40-47. [DOI: https://dx.doi.org/10.1016/j.neuroimage.2007.04.060] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17566764]
Bookheimer, S. Functional MRI of language: new approaches to understanding the cortical organization of semantic processing. Annu Rev Neurosci; 2002; 25, pp. 151-188.[COI: 1:CAS:528:DC%2BD38XmtF2hsLo%3D] [DOI: https://dx.doi.org/10.1146/annurev.neuro.25.112701.142946] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12052907]
Braeutigam, S; Swithenby, SJ; Bailey, AJ. Contextual integration the unusual way: a magnetoencephalographic study of responses to semantic violation in individuals with autism spectrum disorders. Eur J Neurosci; 2008; 27, pp. 1026-1036. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18333970][DOI: https://dx.doi.org/10.1111/j.1460-9568.2008.06064.x]
Brier, MR; Ferree, TC; Maguire, MJ et al. Frontal theta and alpha power and coherence changes are modulated by semantic complexity in Go/NoGo tasks. Int J Psychophysiol; 2010; 78, pp. 215-224. [DOI: https://dx.doi.org/10.1016/j.ijpsycho.2010.07.011] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20696190]
Brock, J; Brown, C; Boucher, J; Rippon, G. The temporal binding deficit hypothesis of autism. Dev Psychopathol; 2002; 14, pp. 209-224. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12030688][DOI: https://dx.doi.org/10.1017/S0954579402002018]
Brouwer, H; Crocker, MW; Venhuizen, NJ; Hoeks, JCJ. A neurocomputational model of the N400 and the P600 in language processing. Cogn Sci; 2017; 41, pp. 1318-1352. [DOI: https://dx.doi.org/10.1111/cogs.12461] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28000963]
Cabeza, R; Nyberg, L. Imaging cognition II: an empirical review of 275 PET and fMRI studies. J Cogn Neurosci; 2000; 12, pp. 1-47.[COI: 1:STN:280:DC%2BD3c3jtl2lug%3D%3D] [DOI: https://dx.doi.org/10.1162/08989290051137585] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10769304]
Caplan, JB; Glaholt, MG; McIntosh, AR. EEG activity underlying successful study of associative and order information. J Cogn Neurosci; 2009; 21, pp. 1346-1364. [DOI: https://dx.doi.org/10.1162/jocn.2008.21167] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19016609]
Catani, M; Mesulam, M. The arcuate fasciculus and the disconnection theme in language and aphasia: History and current state. Cortex; 2008; 44, pp. 953-961. [DOI: https://dx.doi.org/10.1016/j.cortex.2008.04.002] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18614162][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2740371]
Catarino, A; Andrade, A; Churches, O et al. Task-related functional connectivity in autism spectrum conditions: an EEG study using wavelet transform coherence. Mol Autism; 2013; [DOI: https://dx.doi.org/10.1186/2040-2392-4-1] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23311570][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3558480]
Cherkassky, V; Kana, R; Keller, T; Just, M. Functional connectivity in a baseline resting-state network in autism. NeuroReport; 2006; 17, pp. 1687-1690. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17047454][DOI: https://dx.doi.org/10.1097/01.wnr.0000239956.45448.4c]
Coben, R; Clarke, AR; Hudspeth, W; Barry, RJ. EEG power and coherence in autistic spectrum disorder. Clin Neurophysiol; 2008; 119, pp. 1002-1009. [DOI: https://dx.doi.org/10.1016/j.clinph.2008.01.013] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18331812]
Coderre, EL. Dismantling the "visual ease assumption": a review of visual narrative processing in clinical populations. Top Cogn Sci; 2020; 12,
Coderre, EL; Chernenok, M; Gordon, B; Ledoux, K. Linguistic and non-linguistic semantic processing in individuals with autism spectrum disorders: an ERP study. J Autism Dev Disord; 2017; 47, pp. 795-812. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28083778][DOI: https://dx.doi.org/10.1007/s10803-016-2985-0]
Coderre, EL. A semantic priming event-related potential (ERP) task to study lexico-semantic and visuo-semantic processing in autism spectrum disorder. J Vis Exp; 2018; [DOI: https://dx.doi.org/10.3791/57217] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29708550][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5933502]
Delorme, A; Makeig, S. EEGLAB: an open-source toolbox for analysis of single-trial EEG dynamics. J Neurosci Methods; 2004; 134, pp. 9-21. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15102499][DOI: https://dx.doi.org/10.1016/j.jneumeth.2003.10.009]
DiStefano, C; Senturk, D; Jeste, SS. ERP evidence of semantic processing in children with ASD. Dev Cogn Neurosci; 2019; 36, 100640. [DOI: https://dx.doi.org/10.1016/j.dcn.2019.100640] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30974225][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6763343]
Dunn, MA; Bates, JC. Developmental change in neutral processing of words by children with autism. J Autism Dev Disord; 2005; 35, pp. 361-376. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16119477][DOI: https://dx.doi.org/10.1007/s10803-005-3304-3]
Dunn, MA; Gaughan, H, Jr; Kreuzer, J; Kurtzberg, D. Electrophysiologic correlates of semantic classification in autistic and normal children. Dev Neuropsychol; 1999; 16, pp. 79-99. [DOI: https://dx.doi.org/10.1207/S15326942DN160105]
Dunn LM, Dunn DM (2007) PPVT-4: Peabody picture vocabulary test. Pearson Assess
Eichert, N; Verhagen, L; Folloni, D et al. What is special about the human arcuate fasciculus? Lateralization, projections, and expansion. Cortex; 2019; 118, pp. 107-115. [DOI: https://dx.doi.org/10.1016/j.cortex.2018.05.005] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29937266][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6699597]
Fishman, I; Yam, A; Bellugi, U et al. Contrasting patterns of language-associated brain activity in autism and Williams syndrome. Soc Cogn Affect Neurosci; 2011; 6, pp. 630-638. [DOI: https://dx.doi.org/10.1093/scan/nsq075] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20802091]
Fletcher, PT; Whitaker, RT; Tao, R et al. Microstructural connectivity of the arcuate fasciculus in adolescents with high-functioning autism. Neuroimage; 2010; 51, pp. 1117-1125. [DOI: https://dx.doi.org/10.1016/j.neuroimage.2010.01.083] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20132894]
Friederici, AD; Gunter, TC; Hahne, A; Mauth, K. The relative timing of syntactic and semantic processes in sentence comprehension. NeuroReport; 2004; 15, pp. 165-169.[COI: 1:STN:280:DC%2BD2c3gvV2itg%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15106851][DOI: https://dx.doi.org/10.1097/00001756-200401190-00032]
Fries, P. A mechanism for cognitive dynamics: neuronal communication through neuronal coherence. Trends Cogn Sci; 2005; 9, pp. 474-480. [DOI: https://dx.doi.org/10.1016/j.tics.2005.08.011] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16150631]
Frith, U. Autism: explaining the enigma; 1989; Oxford, Blackwell:
Frith, U; Snowling, M. Reading for meaning and reading for sound in autistic and dyslexic children. Br J Dev Psychol; 1983; 1, pp. 329-342. [DOI: https://dx.doi.org/10.1111/j.2044-835X.1983.tb00906.x]
Gaffrey, MS; Kleinhans, NM; Haist, F et al. A typical participation of visual cortex during word processing in autism: an fMRI study of semantic decision. Neuropsychologia; 2007; 45, pp. 1672-1684. [DOI: https://dx.doi.org/10.1016/j.neuropsychologia.2007.01.008] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17336346][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2071933]
Gaffrey, MS; Kleinhans, NM; Haist, F et al. Atypical participation of visual cortex during word processing in autism: an fMRI study of semantic decision. Neuropsychologia; 2008; 45, pp. 1672-1684. [DOI: https://dx.doi.org/10.1016/j.neuropsychologia.2007.01.008]
Glasser, MF; Rilling, JK. DTI Tractography of the human brain’s language pathways. Cereb Cortex; 2008; 18, pp. 2471-2482. [DOI: https://dx.doi.org/10.1093/cercor/bhn011] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18281301]
Groen, Y; Wijers, AA; Mulder, LJM et al. Error and feedback processing in children with ADHD and children with autistic spectrum disorder: an EEG event-related potential study. Clin Neurophysiol; 2008; 119, pp. 2476-2493. [DOI: https://dx.doi.org/10.1016/j.clinph.2008.08.004] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18824404]
Groppe, DM; Makeig, S; Kutas, M. Identifying reliable independent components via split-half comparisons. Neuroimage; 2009; 45, pp. 1199-1211. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19162199][DOI: https://dx.doi.org/10.1016/j.neuroimage.2008.12.038]
Hagoort, P. On Broca, brain, and binding: a new framework. Trends Cogn Sci; 2005; 9, pp. 416-423. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16054419][DOI: https://dx.doi.org/10.1016/j.tics.2005.07.004]
Halgren, E; Kaestner, E; Marinkovic, K et al. Laminar profile of spontaneous and evoked theta: rhythmic modulation of cortical processing during word integration. Neuropsychologia; 2015; 76, pp. 108-124. [DOI: https://dx.doi.org/10.1016/j.neuropsychologia.2015.03.021] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25801916][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4575841]
Happé, FGE. Central coherence and theory of mind in autism: reading homographs in context. Br J Dev Psychol; 1997; 15, pp. 1-12. [DOI: https://dx.doi.org/10.1111/j.2044-835X.1997.tb00721.x]
Harris, GJ; Chabris, CF; Clark, J et al. Brain activation during semantic processing in autism spectrum disorders via functional magnetic resonance imaging. Brain Cogn; 2006; 61, pp. 54-68. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16473449][DOI: https://dx.doi.org/10.1016/j.bandc.2005.12.015]
Holcomb, PJ. Semantic priming and stimulus degradation: Implications for the role of the N400 in language processing. Psychophysiology; 1993; 30, pp. 47-61.[COI: 1:STN:280:DyaK2c%2Fit1Ortw%3D%3D] [DOI: https://dx.doi.org/10.1111/j.1469-8986.1993.tb03204.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/8416062]
Huemer, SV; Mann, V. A comprehensive profile of decoding and comprehension in autism spectrum disorders. J Autism Dev Disord; 2010; 40, pp. 485-493. [DOI: https://dx.doi.org/10.1007/s10803-009-0892-3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19915968]
Hull, JV; Jacokes, ZJ; Torgerson, CM et al. Resting-state functional connectivity in autism spectrum disorders: A review. Front Psychiatry; 2017; [DOI: https://dx.doi.org/10.3389/fpsyt.2016.00205] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28101064][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5209637]
Hus, V; Lord, C. The autism diagnostic observation schedule, module 4: revised algorithm and standardized severity scores. J Autism Dev Disord; 2014; 44, pp. 1996-2012. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24590409][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4104252][DOI: https://dx.doi.org/10.1007/s10803-014-2080-3]
Jackson, RL; Hoffman, P; Pobric, G; Lambon Ralph, MA. The semantic network at work and rest: differential connectivity of anterior temporal lobe subregions. J Neurosci; 2016; 36, pp. 1490-1501.[COI: 1:CAS:528:DC%2BC28Xpt1Ojs7c%3D] [DOI: https://dx.doi.org/10.1523/JNEUROSCI.2999-15.2016] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26843633][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4737765]
Jolliffe, T; Baron-Cohen, S. Linguistic processing in high-functioning adults with autism or Asperger’s syndrome. Is global coherence impaired?. Psychol Med; 2000; 30, pp. 1169-1187.[COI: 1:STN:280:DC%2BD383os1ehsw%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12027052][DOI: https://dx.doi.org/10.1017/S003329179900241X]
Jones, TB; Bandettini, PA; Kenworthy, L et al. Sources of group differences in functional connectivity: an investigation applied to autism spectrum disorder. Neuroimage; 2010; 49, pp. 401-414. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19646533][DOI: https://dx.doi.org/10.1016/j.neuroimage.2009.07.051]
Joseph, RM; Fricker, Z; Fenoglio, A et al. Structural asymmetries of language-related gray and white matter and their relationship to language function in young children with ASD. Brain Imaging Behav; 2014; 8, pp. 60-72. [DOI: https://dx.doi.org/10.1007/s11682-013-9245-0] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23888326]
Just, MA. Cortical activation and synchronization during sentence comprehension in high-functioning autism: evidence of underconnectivity. Brain; 2004; 127, pp. 1811-1821. [DOI: https://dx.doi.org/10.1093/brain/awh199] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15215213]
Just, MA; Cherkassky, VL; Keller, TA; Minshew, NJ. Cortical activation and synchronization during sentence comprehension in high-functioning autism: evidence of underconnectivity. Brain; 2004; 127, pp. 1811-1821. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15215213][DOI: https://dx.doi.org/10.1093/brain/awh199]
Just, MA; Cherkassky, VL; Keller, TA et al. Functional and anatomical cortical underconnectivity in autism: evidence from an fMRI study of an executive function task and corpus callosum morphometry. Cereb Cortex; 2007; 17, pp. 951-961. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16772313][DOI: https://dx.doi.org/10.1093/cercor/bhl006]
Kamio, Y; Toichi, M. Dual access to semantics in autism: is pictorial access superior to verbal access?. J Child Psychol Psychiatry; 2000; 41, pp. 859-867.[COI: 1:STN:280:DC%2BD3M7islGktQ%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/11079428][DOI: https://dx.doi.org/10.1111/1469-7610.00673]
Kamio, Y; Robins, D; Kelley, E et al. Atypical lexical/semantic processing in high-functioning autism spectrum disorders without early language delay. J Autism Dev Disord; 2007; 37, pp. 1116-1122. [DOI: https://dx.doi.org/10.1007/s10803-006-0254-3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17080275]
Kana, RK; Keller, TA; Cherkassky, VL et al. Sentence comprehension in autism: thinking in pictures with decreased functional connectivity. Brain J Neurol; 2006; 129, pp. 2484-2493. [DOI: https://dx.doi.org/10.1093/brain/awl164]
Kaufman AS, Kaufman NL (2004) Kaufman Brief Intelligence Test - Second Edition (KBIT-2). Circle Pines, MN: American Guidance Service
Kelley, E; Paul, JJ; Fein, D; Naigles, LR. Residual language deficits in optimal outcome children with a history of autism. J Autism Dev Disord; 2006; 36, 807. [DOI: https://dx.doi.org/10.1007/s10803-006-0111-4] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16897404]
Klimesch, W; Doppelmayr, M; Pachinger, T; Ripper, B. Brain oscillations and human memory: EEG correlates in the upper alpha and theta band. Neurosci Lett; 1997; 238, pp. 9-12.[COI: 1:CAS:528:DyaK2sXotVyhtLg%3D] [DOI: https://dx.doi.org/10.1016/S0304-3940(97)00771-4] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/9464642]
Koshino, H; Carpenter, PA; Minshew, NJ et al. Functional connectivity in an fMRI working memory task in high-functioning autism. Neuroimage; 2005; 24, pp. 810-821. [DOI: https://dx.doi.org/10.1016/j.neuroimage.2004.09.028] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15652316]
Kulakova, E; Nieuwland, MS. Pragmatic skills predict online counterfactual comprehension: evidence from the N400. Cogn Affect Behav Neurosci; 2016; 16, pp. 814-824. [DOI: https://dx.doi.org/10.3758/s13415-016-0433-4] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27160367][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5018041]
Kutas M, Federmeier KD (2009) N400. Scholarpedia 4:7790. https://doi.org/10.4249/scholarpedia.7790
Kutas, M; Federmeier, KD. Thirty years and counting: finding meaning in the N400 component of the event-related brain potential (ERP). Annu Rev Psychol; 2011; 62, pp. 621-647. [DOI: https://dx.doi.org/10.1146/annurev.psych.093008.131123] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20809790][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4052444]
Lau, EF; Phillips, C; Poeppel, D. A cortical network for semantics: (de)constructing the N400. Nat Rev Neurosci; 2008; 9, pp. 920-933.[COI: 1:CAS:528:DC%2BD1cXhtl2gsb%2FN] [DOI: https://dx.doi.org/10.1038/nrn2532] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19020511]
Lebel, C; Beaulieu, C. Lateralization of the arcuate fasciculus from childhood to adulthood and its relation to cognitive abilities in children. Hum Brain Mapp; 2009; 30, pp. 3563-3573. [DOI: https://dx.doi.org/10.1002/hbm.20779] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19365801][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6870654]
Leocani, L; Toro, C; Manganotti, P et al. Event-related coherence and event-related desynchronization/synchronization in the 10 Hz and 20 Hz EEG during self-paced movements. Electroencephalogr Clin Neurophysiol Potentials Sect; 1997; 104, pp. 199-206.[COI: 1:STN:280:DyaK2szivFOktw%3D%3D] [DOI: https://dx.doi.org/10.1016/S0168-5597(96)96051-7]
López, B; Leekam, SR. Do children with autism fail to process information in context?. J Child Psychol Psychiatry; 2003; 44, pp. 285-300. [DOI: https://dx.doi.org/10.1111/1469-7610.00121] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12587864]
Lopez-Barroso, D; Catani, M; Ripolles, P et al. Word learning is mediated by the left arcuate fasciculus. Proc Natl Acad Sci; 2013; 110, pp. 13168-13173. [DOI: https://dx.doi.org/10.1073/pnas.1301696110] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23884655][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3740909]
Lord, C; Rutter, M; DiLavore, PC et al. Autism diagnostic observation schedule, (ADOS-2) modules 1–4. Kluwer Acad Publ - Plenum Publ; 2012; 30, pp. 204-223.
Maguire, MJ; Abel, AD. What changes in neural oscillations can reveal about developmental cognitive neuroscience: language development as a case in point. Dev Cogn Neurosci; 2013; 6, pp. 125-136. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24060670][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3875138][DOI: https://dx.doi.org/10.1016/j.dcn.2013.08.002]
Manfredi, M; Cohn, N; Sanchez Mello, P et al. Visual and verbal narrative comprehension in children and adolescents with autism spectrum disorders: an ERP study. J Autism Dev Disord; 2020; 50, pp. 2658-2672. [DOI: https://dx.doi.org/10.1007/s10803-020-04374-x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31974801]
Martin, A; Chao, LL. Semantic memory and the brain: structure and processes. Curr Opin Neurobiol; 2001; 11, pp. 194-201.[COI: 1:CAS:528:DC%2BD3MXksFeksbs%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/11301239][DOI: https://dx.doi.org/10.1016/S0959-4388(00)00196-3]
MATLAB. (2010). version 7.10.0 (R2017a). Natick, Massachusetts: The MathWorks Inc
McCleery, JP; Ceponiene, R; Burner, KM et al. Neural correlates of verbal and nonverbal semantic integration in children with autism spectrum disorders. J Child Psychol Psychiatry; 2010; 51, pp. 277-286. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20025622][DOI: https://dx.doi.org/10.1111/j.1469-7610.2009.02157.x]
Mellem, MS; Friedman, RB; Medvedev, AV. Gamma- and theta-band synchronization during semantic priming reflect local and long-range lexical–semantic networks. Brain Lang; 2013; 127, pp. 440-451. [DOI: https://dx.doi.org/10.1016/j.bandl.2013.09.003] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24135132]
Méndez, M; Sans, O; Abril, B; Valdizan, JR. 13. Event-related potentials (N400) in autistic children. Clin Neurophysiol; 2009; 120, e136. [DOI: https://dx.doi.org/10.1016/j.clinph.2008.09.041]
Meyer, L; Grigutsch, M; Schmuck, N et al. Frontal–posterior theta oscillations reflect memory retrieval during sentence comprehension. Cortex; 2015; 71, pp. 205-218. [DOI: https://dx.doi.org/10.1016/j.cortex.2015.06.027] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26233521]
Moseley, RL; Correia, MM; Baron-Cohen, S et al. Reduced volume of the arcuate fasciculus in adults with high-functioning autism spectrum conditions. Front Hum Neurosci; 2016; [DOI: https://dx.doi.org/10.3389/fnhum.2016.00214] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27242478][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4867673]
Mottron, L; Dawson, M; Soulières, I et al. Enhanced perceptual functioning in autism: an update, and eight principles of autistic perception. J Autism Dev Disord; 2006; 36, pp. 27-43. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16453071][DOI: https://dx.doi.org/10.1007/s10803-005-0040-7]
Murias, M; Webb, S; Greenson, J; Dawson, G. Resting state cortical connectivity reflected in EEG coherence in individuals with autism. Biol Psychiatry; 2007; 62, pp. 270-273. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17336944][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2001237][DOI: https://dx.doi.org/10.1016/j.biopsych.2006.11.012]
Nielsen, JA; Zielinski, BA; Fletcher, P et al. Abnormal lateralization of functional connectivity between language and default mode regions in autism. Mol Autism; 2014; 5, 8. [DOI: https://dx.doi.org/10.1186/2040-2392-5-8] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24502324][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3922424]
Nijhof, AD; Dhar, M; Goris, J et al. Atypical neural responding to hearing one’s own name in adults with ASD. J Abnorm Psychol; 2018; 127, pp. 129-138. [DOI: https://dx.doi.org/10.1037/abn0000329] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29369670]
Norbury, C; Nation, K. Understanding variability in reading comprehension in adolescents with autism spectrum disorders: interactions with language status and decoding skill. Sci Stud Read; 2011; 15, pp. 191-210. [DOI: https://dx.doi.org/10.1080/10888431003623553]
O’Rourke E, Coderre EL (2021) Implicit Semantic Processing of Linguistic and Non-linguistic Stimuli in Adults with Autism Spectrum Disorder. J Autism Dev Disord, 51:2611–2630
Ostrolenk, A; Forgeot d’Arc, B; Jelenic, P et al. Hyperlexia: Systematic review, neurocognitive modelling, and outcome. Neurosci Biobehav Rev; 2017; 79, pp. 134-149. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28478182][DOI: https://dx.doi.org/10.1016/j.neubiorev.2017.04.029]
Parellada, M; Penzol, MJ; Pina, L et al. The neurobiology of autism spectrum disorders. Eur Psychiatry; 2014; 29, pp. 11-19.[COI: 1:STN:280:DC%2BC2c3gvFOlsg%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24275633][DOI: https://dx.doi.org/10.1016/j.eurpsy.2013.02.005]
Pijnacker, J; Geurts, B; van Lambalgen, M et al. Exceptions and anomalies: an ERP study on context sensitivity in autism. Neuropsychologia; 2010; 48, pp. 2940-2951. [DOI: https://dx.doi.org/10.1016/j.neuropsychologia.2010.06.003] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20542048]
Price, CJ. The anatomy of language: a review of 100 fMRI studies published in 2009. Ann N Y Acad Sci; 2010; 1191, pp. 62-88. [DOI: https://dx.doi.org/10.1111/j.1749-6632.2010.05444.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20392276]
American Psychiatric Association (2013) Diagnostic and statistical manual of mental disorders, 5th Edition
Rappelsberger, P; Pfurtscheller, G; Filz, O. Calculation of event-related coherence—a new method to study short-lasting coupling between brain areas. Brain Topogr; 1994; 7, pp. 121-127.[COI: 1:STN:280:DyaK2M3htVSksg%3D%3D] [DOI: https://dx.doi.org/10.1007/BF01186770] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/7696089]
Renoult, L; Wang, X; Calcagno, V et al. From N400 to N300: Variations in the timing of semantic processing with repetition. Neuroimage; 2012; 61, pp. 206-215. [DOI: https://dx.doi.org/10.1016/j.neuroimage.2012.02.069] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22406358]
Ribeiro, TC; Valasek, CA; Minati, L; Boggio, PS. Altered semantic integration in autism beyond language: a cross-modal event-related potentials study. NeuroReport; 2013; 24, pp. 414-418. [DOI: https://dx.doi.org/10.1097/WNR.0b013e328361315e] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23629689]
Roach, BJ; Mathalon, DH. Event-related EEG time-frequency analysis: an overview of measures and an analysis of early gamma band phase locking in schizophrenia. Schizophr Bull; 2008; 34, pp. 907-926. [DOI: https://dx.doi.org/10.1093/schbul/sbn093] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18684772][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2632478]
Roberts, TPL; Heiken, K; Zarnow, D et al. Left Hemisphere diffusivity of the arcuate fasciculus: influences of autism spectrum disorder and language impairment. AJNR Am J Neuroradiol; 2014; 35, pp. 587-592.[COI: 1:STN:280:DC%2BC2c3mtlSitw%3D%3D] [DOI: https://dx.doi.org/10.3174/ajnr.A3754] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24335547][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3970288]
Rossion, B; Bodart, J-M; Pourtois, G et al. Functional imaging of visual semantic processing in the human brain. Cortex; 2000; 36, pp. 579-591.[COI: 1:STN:280:DC%2BD3crgslKhtg%3D%3D] [DOI: https://dx.doi.org/10.1016/S0010-9452(08)70539-2] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/11059456]
Sahyoun, CP; Soulières, I; Belliveau, JW et al. Cognitive differences in pictorial reasoning between high-functioning autism and Asperger’s syndrome. J Autism Dev Disord; 2009; 39, pp. 1014-1023. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19267190][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3742807][DOI: https://dx.doi.org/10.1007/s10803-009-0712-9]
Sahyoun, CP; Belliveau, JW; Mody, M. White matter integrity and pictorial reasoning in high-functioning children with autism. Brain Cogn; 2010; 73, pp. 180-188. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20542370][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2905578][DOI: https://dx.doi.org/10.1016/j.bandc.2010.05.002]
Sahyoun, CP; Belliveau, JW; Soulières, I et al. Neuroimaging of the functional and structural networks underlying visuospatial vs. linguistic reasoning in high-functioning autism. Neuropsychologia; 2010; 48, pp. 86-95. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19698726][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2795068][DOI: https://dx.doi.org/10.1016/j.neuropsychologia.2009.08.013]
Samson, F; Mottron, L; Soulières, I; Zeffiro, TA. Enhanced visual functioning in autism: an ALE meta-analysis. Hum Brain Mapp; 2012; 33, pp. 1553-1581. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21465627][DOI: https://dx.doi.org/10.1002/hbm.21307]
Schipul, SE; Keller, TA; Just, MA. Inter-regional brain communication and its disturbance in autism. Front Syst Neurosci; 2011; 5, pp. 1-11. [DOI: https://dx.doi.org/10.3389/fnsys.2011.00010]
Sereno, SC; Rayner, K. Measuring word recognition in reading: eye movements and event-related potentials. Trends Cogn Sci; 2003; 7, pp. 489-493. [DOI: https://dx.doi.org/10.1016/j.tics.2003.09.010] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/14585445]
Shibata, T; Shimoyama, I; Ito, T et al. The synchronization between brain areas under motor inhibition process in humans estimated by event-related EEG coherence. Neurosci Res; 1998; 31, pp. 265-271.[COI: 1:STN:280:DyaK1M%2FisFCguw%3D%3D] [DOI: https://dx.doi.org/10.1016/S0168-0102(98)00046-7] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/9809585]
Siegel, M; Donner, TH; Engel, AK. Spectral fingerprints of large-scale neuronal interactions. Nat Rev Neurosci; 2012; 13, pp. 121-134.[COI: 1:CAS:528:DC%2BC38XlsFyrsA%3D%3D] [DOI: https://dx.doi.org/10.1038/nrn3137] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22233726]
Strandburg, RJ; Marsh, JT; Brown, WS et al. Event-related potentials in high-functioning adult autistics: linguistic and nonlinguistic visual information processing tasks. Neuropsychologia; 1993; 31, pp. 413-434.[COI: 1:STN:280:DyaK3s3nsFGitw%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/8502377][DOI: https://dx.doi.org/10.1016/0028-3932(93)90058-8]
Swank, LK; Catts, HW. Phonological awareness and written word decoding. Lang Speech Hear Serv Sch; 1994; 25, pp. 9-14. [DOI: https://dx.doi.org/10.1044/0161-1461.2501.09]
Tager-Flusberg, H. Semantic processing in the free recall of autistic children: further evidence for a cognitive deficit. Br J Dev Psychol; 1991; 9, pp. 417-430. [DOI: https://dx.doi.org/10.1111/j.2044-835X.1991.tb00886.x]
Tager-Flusberg, H; Paul, R; Lord, C. Volkmar, F; Paul, R; Klin, A; Cohen, D. Language and communication in autism. Handbook of autism and pervasive developmental disorders; 2005; 3 New York, Wiley: pp. 335-364. [DOI: https://dx.doi.org/10.1002/9780470939345.ch12]
Thatcher RW (2004) EEG and Brain Connectivity: A Tutorial. Unpublished manuscript.
Turkeltaub, PE; Flowers, DL; Verbalis, A et al. The neural basis of hyperlexic reading: an fMRI case study. Neuron; 2004; 41, pp. 11-25.[COI: 1:CAS:528:DC%2BD2cXnt12msQ%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/14715131][DOI: https://dx.doi.org/10.1016/S0896-6273(03)00803-1]
Valdizán, JR; Abril-Villalba, B; Méndez-García, M et al. Cognitive evoked potentials in autistic children. Rev Neurol; 2003; 36, pp. 425-428. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/12640594]
Vandenberghe, R; Price, C; Wise, R et al. Functional anatomy of a common semantic system for words and pictures. Nature; 1996; 383, pp. 254-256.[COI: 1:CAS:528:DyaK28Xlsl2ksL0%3D] [DOI: https://dx.doi.org/10.1038/383254a0] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/8805700]
Verbaten, MN; Roelofs, JW; van Engeland, H et al. Abnormal visual event-related potentials of autistic children. J Autism Dev Disord; 1991; 21, pp. 449-470.[COI: 1:STN:280:DyaK387jvVGrug%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/1778960][DOI: https://dx.doi.org/10.1007/BF02206870]
Wan, CY; Marchina, S; Norton, A; Schlaug, G. A typical hemispheric asymmetry in the arcuate fasciculus of completely nonverbal children with autism: Wan et al. Ann N Y Acad Sci; 2012; 1252, pp. 332-337. [DOI: https://dx.doi.org/10.1111/j.1749-6632.2012.06446.x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22524376][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4455878]
Whitehouse, D; Harris, JC. Hyperlexia in infantile autism. J Autism Dev Disord; 1984; 14, pp. 281-289.[COI: 1:STN:280:DyaL2M%2Fgs1Ojug%3D%3D] [DOI: https://dx.doi.org/10.1007/BF02409579] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/6480546]
Yang, J; Small, SL. Wright, JD. Language processing, functional magnetic resonance imaging of. International encyclopedia of the social and behavioral sciences; 2015; 2 Oxford, Elsevier: pp. 368-380. [DOI: https://dx.doi.org/10.1016/B978-0-08-097086-8.54033-1]
Yeatman, JD; Dougherty, RF; Rykhlevskaia, E et al. Anatomical properties of the arcuate fasciculus predict phonological and reading skills in children. J Cogn Neurosci; 2011; 23, pp. 3304-3317. [DOI: https://dx.doi.org/10.1162/jocn_a_00061] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21568636][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3214008]
Zikopoulos, B; Barbas, H. Altered neural connectivity in excitatory and inhibitory cortical circuits in autism. Front Hum Neurosci; 2013; 7, pp. 1-24. [DOI: https://dx.doi.org/10.3389/fnhum.2013.00609]
© The Author(s), under exclusive licence to Springer Nature B.V. 2021.