Biology subjects

Hagoort, P.

Publications and source records attributed to Hagoort, P..

At least 19 recordsLinked to original sources

bioRxiv2026

Transformed reactivation of latent working memory enables hierarchical language processing

Human language comprehension requires tracking words across intervening material to construct grammatical structures, yet two fundamental questions remain unresolved: whether maintenance of earlier words relies on sustained neural activity or activity-silent working memory mechanisms, and whether memory retrieval during integration engages domain-general or language-selective networks. Using magnetoencephalography with time-resolved decoding, we tracked main-clause subjects as participants heard sentences with embedded clauses (e.g., "The dog, who chases the cat, jumps over the mud."). The subject representation ("dog") decayed to baseline during the embedded clause but reactivated after the main-clause verb ("jumps"), with transformed rather than reinstated neural codes. Critically, reactivation emerged first in right dorsolateral prefrontal cortex before engaging frontotemporal language regions, and reactivation strength was modulated by syntactic structure and predicted comprehension accuracy. These findings demonstrate activity-silent working memory maintenance, structure-dependent retrieval, and cooperative function of domain-general and language-selective networks during hierarchical language processing.

neuroscience↗

bioRxiv2026

The scaffolding of individual variability in language processing by domain-general neural networks

Language processing is supported by distributed neural systems. Yet most research examines these systems at the population-average level, obscuring how individual cognitive differences shape language-related brain activity. In this study, we combined comprehensive cognitive assessments and task-based fMRI in a large sample of healthy adults (N = 205) to examine how variability in linguistic knowledge, working memory, processing speed, and non-verbal reasoning influenced neural responses in four language tasks: lexical decision, picture naming, sentence comprehension, and sentence production. All tasks engaged canonical left-lateralized language regions. However, individual differences in cognitive skills were not associated with modulations within commonly activated regions, but rather with modulations in domain-general systems outside traditional perisylvian language areas, mainly the default mode and dorsal attention networks. Notably, activations in these domain-general regions were predominantly negatively correlated with cognitive skills, indicating that individuals with lower cognitive skills draw on these broader neural resources more than higher-skilled individuals, possibly as a compensatory mechanism. These results reveal that while canonical language regions are consistently engaged during language tasks, the recruitment of domain-general systems acts as a variable resource modulated by individuals cognitive skills. Overall, our findings demonstrate that individual cognitive profiles determine how distributed brain systems are dynamically engaged to scaffold language processing.

neuroscience↗

bioRxiv2025

Individual Differences in Speech Monitoring: Functional and Structural Correlates of Delayed Auditory Feedback

Sensory feedback is essential for the fine-tuning of motor actions, and speech production is no exception. It depends on continuous self-monitoring to ensure that produced sounds match intended targets. Delaying auditory feedback (DAF) disrupts this alignment and impairs fluency, providing a powerful tool to investigate sensorimotor control. We combined functional and diffusion-weighted MRI in 31 participants performing a word-production task under delayed (DAF) and immediate (no-DAF) auditory feedback. While all participants slowed their speech under DAF, the extent of this effect varied across individuals and was quantified using a susceptibility index (SI). At the group level, DAF elicited increased activation in a right-lateralized network encompassing the superior temporal gyrus, supramarginal gyrus, inferior frontal gyrus, supplementary motor area, and left cerebellum. Incorporating individual differences revealed that higher susceptibility was associated with greater activation in left-hemisphere speech motor homologues and larger volume of the right long arcuate fasciculus, a white-matter pathway connecting auditory and motor speech regions. This pattern suggests that vulnerability reflects increased recruitment of neural resources and a stronger reliance on auditory-motor coupling. In contrast, resilience was associated with greater engagement of the bilateral angular gyrus and higher fiber density in the right posterior arcuate fasciculus, which connects auditory and somatosensory speech regions. This finding indicates that resilience is supported by a posterior circuit that efficiently integrates multi-modal sensory feedback. Together, these findings link functional dynamics with underlying structural connectivity to reveal how a right-lateralized network supports speech control, while accounting for individual differences in susceptibility to fluency disruption. Significance StatementFluent speech depends on the brains ability to monitor self-produced sounds and sensations from articulatory organs to adjust motor commands in real time. To uncover the neural basis of this process, we combined a fluency-disrupting paradigm, delayed auditory feedback (DAF), with functional and structural neuroimaging. This multimodal approach revealed that while DAF processing relies on a right-lateralized network, more susceptible individuals show enhanced recruitment of left-hemisphere monitoring regions. We also found that stronger white-matter connections in the right posterior arcuate fasciculus predict greater resilience and fluency. These findings provide an anatomically grounded account of how auditory and somatosensory feedback interact to support speech production, offering new insight into why some individuals are more susceptible to fluency breakdowns and related disorders.

neuroscience↗

bioRxiv2024

Neural Encoding of Semantic StructuresDuring Sentence Production

The neural representations for compositional processing have so far been mostly studied during sentence comprehension. In an fMRI study of sentence production, we investigated the brain representations for compositional processing during speaking. We used a rapid serial visual presentation sentence recall paradigm to elicit sentence production from the conceptual memory of an event. With voxel-wise encoding models, we probed the specificity of the compositional structure built during the production of each sentence, comparing an unstructured model of word meaning without relational information with a model that encodes abstract thematic relations and a model encoding event-specific relational structure. Whole-brain analyses revealed that sentence meaning at different levels of specificity was encoded in a large left fronto-parieto-temporal network. A comparison with semantic structures composed during the comprehension of the same sentences showed similarly distributed brain activity patterns. An ROI analysis over left fronto-temporal language parcels showed that event-specific relational structure above word-specific information was encoded in the left inferior frontal gyrus (LIFG). Overall, we found evidence for the encoding of semantic meaning during sentence production in a distributed brain network and for the encoding event-specific semantic structures in the LIFG.

neuroscience↗

bioRxiv2024

Social Anxiety Alters Theory of Mind Activation and Intersubject Neural Variability During Movie Viewing

Social anxiety is characterized by an intense fear of judgment in social situations, yet the underlying mechanisms driving this condition remain poorly understood. One hypothesis holds that specific alterations in Theory of Mind (ToM) affect the ability to interpret others thoughts and emotions. Another hypothesis proposes that broader interpretive biases lead individuals to perceive social cues as overly significant, even in neutral settings. We investigated these possibilities by measuring brain activity, pupil responses, and heart rates in socially anxious individuals and matched controls as they viewed Partly Cloudy, an animated film known to engage the ToM network during specific scenes. While overall brain activity during ToM-related scenes was similar across groups, socially anxious participants exhibited reduced activation in the left posterior superior temporal sulcus (pSTS), a key area for ToM processing. Additionally, intersubject correlation analysis revealed a distinct neural response pattern in the socially anxious group, marked by uniform responses in sensory regions and heightened variability in higher-order cortical areas. This pattern persisted throughout the film and occurred without changes in heart rate or pupil responses, indicating a neural processing bias that manifests even in non-evaluative settings. These findings provide a neural basis for ToM alterations and broader interpretive biases in social anxiety, supporting cognitive-behavioral models and suggesting novel targets for intervention.

neuroscience↗

bioRxiv2024

Susceptibility to auditory feedback manipulations and individual variability

Monitoring auditory feedback from hearing ones own voice is important for fluent speech production as it enables detection and correction of speech errors. The influence of auditory feedback is best illustrated by manipulating it during speech production. A common temporal manipulation technique, delaying auditory feedback (DAF), leads to disruptions in speech fluency, while a common spectral manipulation technique, perturbing the pitch of auditory feedback (PAF), results in vocal alterations. Previous research involving clinical populations has revealed diverse susceptibility profiles to auditory feedback manipulations, yet the extent of such diversity within the neurotypical population remains unclear. Furthermore, different types of manipulations elicit distinct speech errors (i.e. fluency/coordination versus acoustic errors), which may be processed by distinct error correction mechanisms. It is yet to be understood whether individuals affected by one manipulation are similarly impacted by the other. Lastly, based on evidence from clinical studies, which demonstrated that visual feedback can improve impaired speech production, it is an open question whether visual feedback can alleviate the disruptive effects of altered auditory feedback. We recorded voice samples from 40 neurotypical participants during both a DAF and a PAF task. DAF significantly prolonged articulation duration and increased voice pitch and intensity. In some trials, participants received immediate visual feedback, however visual feedback did not alleviate but rather strengthened the disruptive effects of DAF. During the PAF task, participants adjusted their voice pitch in the opposite direction of the perturbation in majority of the trials to compensate for the perturbation. We assessed susceptibility of the participants to the effects of DAF and PAF by examining articulation duration and compensatory vocal response magnitude, respectively. Susceptibility varied widely among participants for both manipulations, but individuals susceptible to one manipulation did not consistently exhibit susceptibility to the other, indicating distinct processing mechanisms for these different types of auditory feedback manipulations.

neuroscience↗

bioRxiv2024

Nonlinear dendritic integration supports Up-Down states in single neurons

Changes in the activity profile of cortical neurons are due to effects at the scale of local and long-range networks. Accordingly, abrupt transitions in the state of cortical neurons--a phenomenon known as Up-Down states--have been attributed to variation in the activity of afferent neurons. However, cellular physiology and morphology may also play a role in causing Up-Down states. This study examines the impact of dendritic nonlinearities, particularly those mediated by voltage-dependent NMDA receptors, on the response of cortical neurons to balanced excitatory/inhibitory synaptic inputs. Using a neuron model with two segregated dendritic compartments, we compared cells with and without dendritic nonlinearities. NMDA receptors boosted somatic firing in the balanced condition and increased the correlation between membrane potentials across the compartments of the neuron model. Dendritic nonlinearities elicited strong bimodality in the distribution of the somatic potential when the cell was driven with cortical-like input. Moreover, dendritic nonlinearities could detect small input fluctuations and lead to Up-Down states whose statistics and dynamics closely resemble electrophysiological data. Up-Down states also occurred in recurrent networks with oscillatory firing activity, as in anaesthetized animal models, when dendritic NMDA receptors were partially disabled. These findings suggest that there is a dissociation between cellular and network-level features that could both contribute to the emergence of Up-Down states. Our study highlights the complex interplay between dendritic integration and activity-driven dynamics in the origin of cortical bistability. Significance statementIn several physiological states, such as sleep or quiet wakefulness, the membrane of cortical cells shows characteristic bistability. Cells are either fully depolarized and ready to spike, or in a silent, hyperpolarized state. This dynamics, known as Up-Down states, has often been attributed to changes in network activity. However, whether cell-specific properties, such as dendritic nonlinearities, play a role in neuronal bistability remains unclear. This study uses a dendritic model of a pyramidal cell and shows that the presence of NMDA receptors drives the Up-Down states in response to small fluctuations in network activity. Thus, single cells can enter the Up-Down state dynamics independently of the ongoing network activity.

neuroscience↗

bioRxiv2024

Neural dynamics express syntax in the time domain during natural story listening

Studies of perception have long shown that the brain adds information to its sensory analysis of the physical environment. A touchstone example for humans is language use: to comprehend a physical signal like speech, the brain must add linguistic knowledge, including syntax. Yet, syntactic rules and representations are atemporal (i.e., abstract and not bound by time), so they must be translated into time-varying signals for speech comprehension and production. Here, we test three different models of the temporal spell-out of syntactic structure against brain activity of people listening to Dutch stories: an integratory bottom-up parser, a predictive top-down parser, and a mildly predictive left-corner parser. These models build exactly the same structure but differ in when syntactic information is added by the brain - this difference is captured in the (temporal distribution of the) complexity metric incremental node count. Using temporal response function models with both acoustic and information-theoretic control predictors, node counts were regressed against source-reconstructed delta-band activity acquired with magnetoencephalography. Neural dynamics in left frontal and temporal regions most strongly reflect node counts derived by the top-down method, which postulates syntax early in time, suggesting that predictive structure building is an important component of Dutch sentence comprehension. The absence of strong effects of the left-corner model further suggests that its mildly predictive strategy does not represent Dutch language comprehension well, in contrast to what has been found for English. Understanding when the brain projects its knowledge of syntax onto speech, and whether this is done in language-specific ways, will inform and constrain the development of mechanistic models of syntactic-structure building in the brain.

neuroscience↗

bioRxiv2024

Preserved spontaneous mentalizing amid reduced intersubject variability in autism during a movie narrative

Individuals with autism may perform well in structured Theory of Mind (ToM) tasks that assess their ability to infer mental states, yet encounter challenges in everyday social interactions. Using fMRI and pupillometry, we investigated whether this discrepancy stems from reduced spontaneous mentalizing or broader difficulties in unstructured environments. Autistic adults and neurotypical controls viewed a nonverbal animated movie featuring events known to evoke mental state inferences. Both groups exhibited similar ToM network activations and pupil responses to these events, alongside comparable verbal accounts of the mental states portrayed. However, intersubject correlation analysis revealed a significant reduction in response variability among autistic participants throughout the movies complex visual narrative, affecting brain regions beyond the ToM network. We suggest that the primary challenge for individuals with autism may lie in the idiosyncratic exploration of narratives in unstructured settings, regardless of whether mental state inferences are involved.

neuroscience↗

bioRxiv2023

Dendrites support formation and reactivation of sequential memories through Hebbian plasticity

Storage and retrieval of sequences require memory that is sensitive to the temporal order of features. For example, in human language, words that are stored in long-term memory are retrieved based on the order of phonemes. It is currently unknown whether Hebbian learning supports the formation of memories that are structured in time. We investigated whether word-like memories can emerge in a network of neurons with dendritic structures. Dendrites provide neuronal processing memory on the order of 100 ms and have been implicated in structured memory formation. We compared a network of neurons with dendrites and two networks of point neurons that have previously been shown to acquire stable long-term memories and process sequential information. The networks were equipped with voltage-based, spike-timing dependent plasticity (STDP) and were homeostatically balanced with inhibitory STDP. In the learning phase, networks were exposed to phoneme sequences and word labels, which led to the formation of overlapping cell assemblies. In the retrieval phase, networks only received phoneme sequences as input, and we measured the firing activity of the corresponding word populations. The dendritic network correctly reactivated the word populations with a success rate of 80%, including words composed of the same phonemes in a different order. The networks of point neurons reactivated only words that contained phonemes that were unique to these words and confused words with shared phonemes (success rate below 20%). These results suggest that the slow timescale and non-linearity of dendritic depolarization allowed neurons to establish connections between neural groups that were sensitive to serial order. Inhibitory STDP prevented the potentiation of connections between unrelated neural populations during learning. During retrieval, it maintained the dendrites hyperpolarized and limited the reactivation of incorrect cell assemblies. Thus, the addition of dendrites enables the encoding of temporal relations into associative memories.

neuroscience↗

bioRxiv2023

Attention drives visual processing and audiovisual integration during multimodal communication

During communication in real-life settings, our brain often needs to integrate auditory and visual information, and at the same time actively focus on the relevant sources of information, while ignoring interference from irrelevant events. The interaction between integration and attention processes remains poorly understood. Here, we use rapid invisible frequency tagging (RIFT) and magnetoencephalography (MEG) to investigate how attention affects auditory and visual information processing and integration, during multimodal communication. We presented human participants (male and female) with videos of an actress uttering action verbs (auditory; tagged at 58 Hz) accompanied by two movie clips of hand gestures on both sides of fixation (attended stimulus tagged at 65 Hz; unattended stimulus tagged at 63 Hz). Integration difficulty was manipulated by a lower-order auditory factor (clear/degraded speech) and a higher-order visual semantic factor (matching/mismatching gesture). We observed an enhanced neural response to the attended visual information during degraded speech compared to clear speech. For the unattended information, the neural response to mismatching gestures was enhanced compared to matching gestures. Furthermore, signal power at the intermodulation frequencies of the frequency tags, indexing non-linear signal interactions, was enhanced in left frontotemporal and frontal regions. Focusing on LIFG, this enhancement was specific for the attended information, for those trials that benefitted from integration with a matching gesture. Higher power at this intermodulation frequency was related to faster reaction times. Together, our results suggest that attention modulates the strength and speed of audiovisual processing and interaction, depending on the congruence and quality of the sensory input.

neuroscience↗

bioRxiv2022

Diverging Neural Dynamics for Syntactic Structure Building in Naturalistic Speaking and Listening

The neural correlates of sentence production have been mostly studied with constraining task paradigms that introduce artificial task effects. In this study, we aimed to gain a better understanding of syntactic processing in spontaneous production vs. naturalistic comprehension. We extracted word-by-word metrics of phrase-structure building with top-down and bottom-up parsers that make different hypotheses about the timing of structure building. In comprehension, structure building proceeded in an integratory fashion and led to an increase in activity in posterior temporal and inferior frontal areas. In production, structure building was anticipatory and predicted an increase in activity in the inferior frontal gyrus. Newly developed production-specific parsers highlighted the anticipatory and incremental nature of structure building in production, which was confirmed by a converging analysis of the pausing patterns in speech. Overall, the results showed that the unfolding of syntactic processing diverges between speaking and listening.

neuroscience↗

bioRxiv2022

The Interaction of Context Constraints and Predictive Validity during Sentence Reading

Words are not processed in isolation, instead they are commonly embedded in phrases and sentences. The sentential context influences the perception and processing of a word. However, how this is achieved by brain processes and whether predictive mechanisms underlie this process remains a debated topic. To this end we employed an experimental paradigm in which we orthogonalized sentence context constraints and predictive validity, which was defined as the ratio of congruent to incongruent sentence endings within the experiment. While recording electroencephalography, participants read sentences with three levels of sentential context constraints (high, medium and low). Participants were also separated into two groups, which differed in their ratio of valid congruent to incongruent target words that could be predicted from the sentential context. For both groups we investigated modulations of alpha power before, and N400 amplitude modulations after target word onset. The results reveal that the N400 amplitude gradually decreases with higher context constraints. Contrary, alpha power is non-monotonically influenced, displaying the strongest decrease for high context constraints over frontal electrode sites, while alpha power between medium and low context constraints does not differ. This indicates that both neural correlates are influenced by the degree of context constraint but are not affected by changes in predictive validity. The results therefore suggest that both N400 and alpha power are not unequivocally linked to the predictability of a target word based on larger contextual information.

neuroscience↗

bioRxiv2022

The Tripod neuron: a minimal structural reduction of the dendritic tree

Neuron models with explicit dendritic dynamics have shed light on mechanisms for coincidence detection, pathway selection, and temporal filtering. However, it is still unclear which morphological and physiological features are required to capture these phenomena. In this work, we introduce the Tripod neuron model and propose a minimal structural reduction of the dendritic tree that is able to reproduce these dendritic computations. The Tripod is a three-compartment model consisting of two segregated passive dendrites and a somatic compartment modeled as an adaptive, exponential integrate-and-fire neuron. It incorporates dendritic geometry, membrane physiology, and receptor dynamics as measured in human pyramidal cells. We characterize the response of the Tripod to glutamatergic and GABAergic inputs and identify parameters that support supra-linear integration, coincidence-detection, and pathway-specific gating through shunting inhibition. Following NMDA spikes, the Tripod neuron generates plateau potentials whose duration depends on the dendritic length and the strength of synaptic input. When fitted with distal compartments, the Tripod neuron encodes previous activity into a dendritic depolarized state. This dendritic memory allows the neuron to perform temporal binding and we show that the neuron solves transition and sequence detection tasks on which a single-compartment model fails. Thus, the Tripod neuron can account for dendritic computations previously explained only with more detailed neuron models or neural networks. Due to its simplicity, the Tripod model can be used efficiently in simulations of larger cortical circuits.

neuroscience↗

bioRxiv2022

Neural tracking of speech does not unequivocally reflect intelligibility

During listening, brain activity tracks the rhythmic structures of speech signals. Here, we directly dissociated the contribution of neural tracking in the processing of speech acoustic cues from that related to linguistic processing. We examined the neural changes associated with the comprehension of Noise-Vocoded (NV) speech using magnetoencephalography (MEG). Participants listened to NV sentences in a 3-phase training paradigm: (1) pre-training, where NV stimuli were barely comprehended, (2) training with exposure of the original clear version of speech stimulus, and (3) post-training, where the same stimuli gained intelligibility from the training phase. Using this paradigm, we tested if the neural responses of a speech signal was modulated by its intelligibility without any change in its acoustic structure. To test the influence of spectral degradation on neural tracking independently of training, participants listened to two types of NV sentences (4-band and 2-band NV speech), but were only trained to understand 4-band NV speech. Significant changes in neural tracking were observed in the delta range in relation to the acoustic degradation of speech. However, we failed to find a direct effect of intelligibility on the neural tracking of speech in both theta and delta ranges. This suggests that acoustics greatly influence the neural tracking response to speech signals, and that caution needs to be taken when choosing the control signals for speech-brain tracking analyses, considering that a slight change in acoustic parameters can have strong effects on the neural tracking response.

neuroscience↗

bioRxiv2021

Prediction and preview strongly affect reading times but not skipping during natural reading

In a typical text, readers look much longer at some words than at others and fixate some words multiple times, while skipping others altogether. Historically, researchers explained this variation via low-level visual or oculomotor factors, but today it is primarily explained in terms of cognitive factors, such as how well word identity can be predicted from context or discerned from parafoveal preview. While the existence of these effects has been well established in experiments, the relative importance of prediction, preview and low-level factors for eye movement variation in natural reading is unclear. Here, we address this question in three large datasets (n=104, 1.5 million words), using a deep neural network and Bayesian ideal observer to model linguistic prediction and parafoveal preview from moment to moment in natural reading. Strikingly, neither prediction nor preview was important for explaining word skipping - the vast majority of skipping was explained by a simple oculomotor model. For reading times, by contrast, we found strong but independent contributions of both prediction and preview, with effect sizes matching those from controlled experiments. Together, these results challenge dominant models of eye movements in reading by showing that linguistic prediction and parafoveal preview are not important determinants of word skipping.

neuroscience↗

bioRxiv2021

Minimal phrase composition revealed by intracranial recordings

The ability to comprehend phrases is an essential integrative property of the brain. Here we evaluate the neural processes that enable the transition from single word processing to a minimal compositional scheme. Previous research has reported conflicting timing effects of composition, and disagreement persists with respect to inferior frontal and posterior temporal contributions. To address these issues, 19 patients (10 male, 19 female) implanted with penetrating depth or surface subdural intracranial electrodes heard auditory recordings of adjective-noun, pseudoword-noun and adjective-pseudoword phrases and judged whether the phrase matched a picture. Stimulus-dependent alterations in broadband gamma activity, low frequency power and phase-locking values across the language-dominant left hemisphere were derived. This revealed a mosaic located in the posterior superior temporal sulcus (pSTS), in which closely neighboring cortical sites displayed exclusive sensitivity to either lexicality or phrase structure, but not both. Distinct timings were found for effects of phrase composition (210-300 ms) and pseudoword processing (approximately 300-700 ms), and these were localized to neighboring electrodes in pSTS. The pars triangularis and temporal pole encode anticipation of composition in broadband low frequencies, and both regions exhibit greater functional connectivity with pSTS during phrase composition. Our results suggest that the pSTS is a highly specialized region comprised of sparsely interwoven heterogeneous constituents that encodes both lower and higher level linguistic features. This hub in pSTS for minimal phrase processing may form the neural basis for the human-specific computational capacity for forming hierarchically organized linguistic structures. SignificanceLinguists have claimed that the integration of multiple words into a phrase demands a computational procedure distinct from single word processing. Here, we provide intracranial recordings from a large patient cohort, with high spatiotemporal resolution, to track the cortical dynamics of phrase composition. Epileptic patients volunteered to participate in a task in which they listened to phrases ("red boat"), word-pseudoword or pseudoword-word pairs (e.g., "red fulg"). At the onset of the second word in phrases, greater broadband high gamma activity was found in posterior superior temporal sulcus in electrodes that exclusively indexed phrasal meaning, and not lexical meaning. These results provide direct, high-resolution signatures of minimal phrase composition in humans, a potentially species-specific computational capacity.

neuroscience↗

bioRxiv2021

MVPA does not reveal neural representations of hierarchical linguistic structure in MEG

During comprehension, the meaning extracted from serial language input can be described by hierarchical phrase structure. Whether our brains explicitly encode hierarchical structure during processing is, however, debated. In this study we recorded Magnetoencephalography (MEG) during reading of structurally ambiguous sentences to probe neural activity for representations of underlying phrase structure. 10 human subjects were presented with simple sentences, each containing a prepositional phrase that was ambiguous with respect to its attachment site. Disambiguation was possible based on semantic information. We applied multivariate pattern analyses (MVPA) to the MEG data using linear classifiers as well as representational similarity analysis to probe various effects of phrase structure building on the neural signal. Using MVPA techniques we successfully decoded both syntactic (part-of-speech) as well as semantic information from the brain signal. Importantly, however, we did not find any patterns in the neural signal that differentiate between different hierarchical structures. Nor did we find neural traces of syntactic or semantic reactivation following disambiguating sentence material. These null findings suggest that subjects may not have processed the sentences with respect to their underlying phrase structure. We discuss methodological limits of our analysis as well as cognitive theories of "shallow processing", i.e. in how far rich semantic information can prevent thorough syntactic analysis during processing.

neuroscience↗