Language Network Research Articles

Neural network mechanisms of emotional dysregulation in cerebral small vessel disease.

Evidence of language network reorganization and compromised cognitive functioning in pediatric patients with focal refractory epilepsy.

Abstract Lewy body diseases, including Parkinson’s disease and dementia with Lewy bodies, often involve mild cognitive impairment at diagnosis (mild cognitive impairment with Lewy bodies (MCI-LB). Language dysfunction in MCI-LB patients is often unrecognized. This study aimed to assess syntactic comprehension deficits in MCI-LB patients and to explore their neural correlates. A total of 25 MCI-LB patients (mean±sd: 72±5.6 years old, 10 women) and 25 healthy controls (HC, mean±sd: 66±4.0 years old, 12 women) performed task functional MRI Test of Sentence Comprehension (ToSC). Functional connectivity was analyzed using psychophysiological interaction (PPI) method, focusing on the striatum and language networks. MCI-LB patients had lower ToSC scores than HC (MCI-LB: 74.7±15.7, HC: 88.5±9.0, p<0.001) and their PPI analysis revealed decreased connectivity from the striatum to the cuneus, precuneus, and left supramarginal gyrus, and reduced connectivity particularly in the dorsal pathway during noncanonical (syntactically more complex) sentence processing. Taken together, in this cross-sectional study MCI-LB patients showed impaired sentence comprehension related to decreased subcortical-cortical and dorsal language network connectivity. Specific changes in frontotemporal connectivity in MCI-LB might be a promising indicator of language related cognitive impairment in these a-synucleinopathies.

IntroductionEpilepsy due to hypothalamic hamartoma (HH) is associated with epileptic encephalopathy and often requires surgical intervention, as medications are ineffective at reducing the seizures. However, the first step of disentangling the impact of the surgery on the broader whole-brain networks, a biomarker of encephalopathy compared to controls, is not quantified. Subtle pre- and post-operative networks can elude conventional rs-fMRI analysis.MethodsWe retrospectively analyzed rs-fMRI from 56 HH patients scanned before and 6 months after surgery. We developed a two-stage contrastive learning-based algorithm to classify the motor, vision, language, frontal, and temporal networks as pre- vs post-operative. In stage one, a multimodal contrastive encoder jointly ingests 3D spatial Independent Component Analysis (ICA) maps and their corresponding 1D temporal ICA time series to learn embeddings that distinguish pre-operative from post-operative states for each network while separating embeddings of different networks. In stage two, a lightweight classifier refines these embeddings, augmented by original ICA inputs, to classify each network as pre-operative or post-operative.ResultsVisualization of the learned feature space with t-SNE revealed clear separation by pre- vs post-surgical condition across all five networks. Across networks, mean accuracy ranged from 0.85 to 0.90, sensitivity from 0.79 to 0.90, specificity from 0.87 to 0.93, F1‐score from 0.83 to 0.90 and AUC from 0.90 to 0.94 in stratified cross validation.ConclusionsContrastive learning can sensitively detect functional shifts in critical cortical networks that previous traditional analyses may overlook. These findings inform broader shifts in whole-brain network functioning following effective HH surgery and establish a featurewise distinction between preoperative and postoperative states, motivating future studies that compare HH patients to healthy controls to quantify network recovery.

In recent advancements within natural language processing (NLP), lexical networks play a crucial role in representing semantic relationships between words, enhancing applications from word sense disambiguation to educational tools. Traditional methods for constructing lexical networks, however, are resource-intensive, relying heavily on expert lexicographers. Leveraging GPT-4o, a large language model (LLM), our study presents an automated, scalable approach to creating multi-relational Japanese lexical networks for the general Japanese language. This study builds on previous methods of integrating synonyms but extends to other relations such as hyponymy, hypernymy, meronymy, and holonomy. Using a combination of structured prompts and graph-based data storage, the model extracts detailed lexical relationships, which are then systematically validated and encoded. Results reveal a substantial expansion in network size, with over 155,000 nodes and 700,000 edges, enriching Japanese lexical associations with nuanced hierarchical and associative layers. Comparisons with WordNet show substantial alignment in relation types, particularly with soft matching, underscoring the model’s efficacy in reflecting the multifaceted nature of lexical semantics. This work contributes a versatile framework for constructing expansive lexical resources that hold promises for enhancing NLP tasks and educational applications across various languages and domains.

BackgroundSymptoms of attention-deficit/hyperactivity disorder (ADHD) are common in children with autism spectrum disorder (ASD), and are associated with greater developmental challenges, poorer clinical outcomes, and alterations in functional connectivity (FC) of the brain. However, despite the consensus that ASD and other neurodevelopmental conditions emerge early in life, little is known about the trajectories of brain and behavioral development during the first years of life in children with ASD and co-occurring attention problems (AP).MethodsIn a sample of 122 young children (ages 1.5–5 years) with and without ASD, we examined whether toddlers and preschoolers with ASD and co-occurring AP already differ from peers with ASD without co-occurring AP on adaptive and developmental skills, ASD symptoms, and FC of the frontoparietal and salience networks, which have been previously linked to ADHD symptoms in older children with ASD and ADHD.ResultsResults of general linear model analyses revealed lower developmental and adaptive skills across multiple domains in children with ASD and elevated AP compared with their peers with lower AP, despite equivalent levels of ASD symptoms. Further, children with ASD and elevated AP showed reduced FC within the frontoparietal network (p = .027), between the frontoparietal and language networks (p = .004), and the frontoparietal and default mode networks (p = .046) in comparison to their peers with lower AP. No group differences in FC of the salience network were observed (all p > .05).ConclusionsThese findings provide evidence that neurodevelopmental and behavioral differences in children with ASD and co-occurring AP emerge very early in life, before a reliable diagnosis of ADHD is typically made. Specifically, these results demonstrate that early inattention symptoms are associated with unique connectivity patterns in executive circuitry as early as the first years of life in toddlers and preschoolers with ASD, likely contributing to the phenotypic and neural heterogeneity recognized in autism. Thus, our results underscore the importance of considering co-occurring conditions early in developmental research and clinical care, as further understanding these trajectories can inform early interventions during the critical time period when they have the greatest potential for positive impact.

Research on how functional connectivity between networks changes during motor performance is limited. Therefore, this study analyzed the effects of elbow flexion and extension without loading on the functional connectivity between brain networks in 15 healthy adult males. The experimental design consisted of a rest phase (1 min) and a motion phase (1[Formula: see text]min). Brain activity data were collected using functional magnetic resonance imaging (fMRI), and the SPM12 and CONN toolboxes were used to analyze functional connectivity. Connectivity between the sensorimotor and dorsal attention networks increased during motor performance, whereas connectivity with the default mode network (DMN) decreased. In addition, a new connectivity between the visual and language networks was formed. Moreover, while connectivity with the anterior insula in the salience network was found to increase, connectivity with the anterior cingulate cortex (ACC) decreased. These findings suggest that motor performance is not confined to simple sensorimotor activity; it is regulated by interactions with the attention, visual and salience networks. This study analyzed the changes in coordination between networks during motor performance from the perspective of functional connectivity. We expect these findings to be useful neurological resources for further research on motor rehabilitation and learning.

Test-retest reliability of edge-level resting-state functional connectivity in people with aphasia

Traditional emotion decoding methods typically rely on short sequences with limited context and coarse-grained emotion categories. To address these limitations, we proposed the Semantic and Emotion Decoding Generative Pre-trained Transformer (SED-GPT), a non-invasive method for long-sequence fine-grained semantics and emotions decoding on extended narrative stimuli. Using a publicly available fMRI dataset from 8 participants, this exploratory study investigates the feasibility of reconstructing complex semantic and emotional states from brain activity. SED-GPT achieves a BERTScore-F1 of 0.650 on semantic decoding and attains a cosine similarity (CS) of 0.504 and a Jensen–Shannon similarity (JSS) of 0.469 for emotion decoding (p < 0.05). Functional connectivity analyses reveal persistent coupling between the language network and the emotion network, which provides neural evidence for the language–emotion interaction mechanism in Chinese. These findings should be interpreted as pilot-level feasibility evidence.

In this two-part investigation, we examined whether Alzheimer's disease (AD) phenotypes are distinct clinical entities or represent positions within a graded multidimensional space. First, using a large retrospective dataset of past research participants (n = 413) from memory clinics, we examined the comparative distributions of cognitive performance in people diagnosed with typical amnestic AD (tAD), logopenic variant of primary progressive aphasia (lvPPA), and posterior cortical atrophy (PCA), across a broad range of disease severities. Secondly, a prospective deep phenotyping study of lvPPA (n = 18) compared to typical AD (n = 9) addressed the following questions: (1) Does the multidimensional cognitive pattern of impairment only emerge in advanced lvPPA, and how does it compare to tAD? (2) Do memory deficits in lvPPA appear in a simple clinic-level cognitive assessment or require in-depth neuropsychological investigation? (3) To what extent is performance on verbal episodic memory attributable to language impairment? (4) Do the patterns of decline in lvPPA and tAD stay categorical or multidimensional over time? We explored the associations between scores derived from a principal component analysis of cognitive measures, and grey matter volumes in key memory- and language-related brain regions, at baseline and longitudinally. The clinic-level assessment revealed similar results in both the prospective and retrospective data: (i) patients showed graded distinctions (e.g., predominant visual versus language impairment in people with PCA versus lvPPA) and overlap (e.g., shared weakness in domains such as memory); and (ii) people with lvPPA and tAD were equally impaired on both verbal and non-verbal memory tests. Longitudinal assessment showed phenotypic dispersion: (i) people with tAD showed varied patterns of phenotypic differentiation; and (ii) people with lvPPA and lvPPA + exhibited a multidimensional pattern of decline with decreasing principal component scores and worsening multi-domain cognitive performance. The results of Bayesian linear regressions showed evidence for the association of grey matter volumes in language and memory networks with principal component analysis derived scores. The graded distinctions amongst typical amnestic and atypical (language and visual) phenotypes of AD support the proposal for a transdiagnostic, multidimensional phenotype geometry that spans all AD subtypes.

We introduce a general approach for analyzing large-scale text-based data, combining the strengths of neural network language processing and generative statistical modeling to create a factor structure of unstructured data for downstream regressions typically used in social sciences. We generate textual factors by (i) representing texts using vector word embedding, (ii) clustering the vectors using locality-sensitive hashing to generate supports of topics, and (iii) identifying relatively interpretable spanning clusters (i.e., textual factors) through topic modeling. Our data-driven approach captures complex linguistic structures while ensuring computational scalability and economic interpretability, plausibly attaining certain advantages over and complementing other unstructured data analytics used by researchers, including emergent large language models. We conduct initial validation tests of the framework and discuss three types of its applications: (i) enhancing prediction and inference with texts, (ii) interpreting (non–text-based) models, and (iii) constructing new text-based metrics and explanatory variables. We illustrate each of these applications using examples in finance and economics such as macroeconomic forecasting from news articles, interpreting multifactor asset pricing models from corporate filings, and measuring theme-based technology breakthroughs from patents. Finally, we provide a flexible statistical package of textual factors for online distribution to facilitate future research and applications. This paper was accepted by David Simchi-Levi, finance. Funding: The authors gratefully acknowledge the financial support from the Ewing Marion Kauffman Foundation, the Becker Friedman Institute of Economics, the Fama-Miller Center for Research in Finance, INQUIRE Europe, the Kenan Institute of Private Enterprise, and the Risk Institute at OSU Fisher College of Business (while L. W. Cong was a fellow at the institute). W. Zhu acknowledges financial support from the Tsinghua University Initiative Scientific Research Program [Grant 2022Z04W02016], the Tsinghua University School of Economics and Management [Research Grant 2022051002], and the National Natural Science Foundation of China [Grant 72442014]. Supplemental Material: The online appendices and data files are available at https://doi.org/10.1287/mnsc.2020.01180 .

This study investigated dynamic cerebellar networks in post-stroke aphasia patients using resting-state fMRI. We examined intra-cerebellar and cerebellar-cortical dynamic functional connectivity quantified their temporal properties and graph-theoretical topology. Seventy-seven right-handed patients with post-stroke aphasia and 79 healthy controls underwent underwent 3T resting-state functional MRI. Dynamic cerebellar functional networks were constructed using the Seitzman-27 cerebellar atlas. A sliding-window approach (30 TR window, 1 TR step) was applied, followed by k-means clustering to identify distinct connectivity states. Graph-theoretical analyses were performed to quantify state-specific network topology. Variability of dynamic functional connectivity between cerebellar and cortical regions was calculated. Partial correlation analyses were conducted to examine relationships between dynamic network measures, lesion volume, and language and cognitive function. Two cerebellar dynamic functional connectivity states were identified in post-stroke aphasia: a predominant segregated state (78.93%) with widespread reductions in connectivity and decreased clustering coefficient (d = -1.29), characteristic path length (d = -0.62), and Local Efficiency (d = -1.11), but higher Global Efficiency (d = 1.06); and a less frequent integrated state (21.07%) with enhanced connectivity and higher Clustering Coefficient (d = 0.57) and Characteristic Path Length (d = 0.70), but diminished Global Efficiency (d = -1.25) and small-worldness (d = -0.92), small-world index (d = -0.89). Post-stroke aphasia showed reduced variability of dynamic functional connectivity between cerebellar and cortical regions involved in language and cognition (Gaussian random field correction, voxel-level p < 0.001, cluster-level p < 0.05). Lesion volume negatively correlated with Aphasia Quotient, Repetition, Memory, Executive Function, and Attention (p < 0.05). State-specific network metrics and variability measures were associated with language and cognitive performance independently of lesion volume. Post-stroke aphasia patients exhibited a segregated cerebellar state with reduced intra-cerebellar connectivity and efficiency, and an integrated state with enhanced connectivity and small-world properties, together with reduced variability in cerebellar-cortical connections to language-and cognition-related regions. These state-specific network alterations were linked to distinct behavioral domains independently of lesion volume, highlighting a dissociation between structural constraints and dynamic, lesion-independent plasticity. PSA = Post-Stroke Aphasia; DFC = Dynamic Functional Connectivity; FPN = Frontoparietal Network; DMN = Default Mode Network; SMN = Somatomotor Network; Eloc = Local Efficiency; Eg = Global Efficiency; Lp = Characteristic Path Length; DAN = Dorsal Attention Network; FDR = False Discovery Rate.

Interhemispheric asymmetry is a core feature of human brain organization, yet its functional relevance across cognitive tasks remains incompletely understood. Using data from 989 healthy adults, we examined patterns of functional asymmetry and their relationship to bilateral fMRI signal amplitude and task performance across seven tasks: motor, language, social cognition, relational processing, working memory, gambling, and emotion. An fMRI-derived asymmetry index was computed across 17 task epochs and mapped onto the cortical surface. Here we show that both fMRI signal amplitude and asymmetry were positively associated with task accuracy across multiple networks and tasks epochs. These associations were strongest in language, frontoparietal, and dorsal attention networks during high-demand tasks, such as story comprehension, relational processing, and working memory. Partial least squares regression revealed that amplitude was a more robust predictor of task accuracy than asymmetry. These findings suggest that greater neural activation drives stronger hemispheric differentiation and supports cognitive performance.

The ability to generalize previously learned knowledge to novel situations is crucial for adaptive behavior, representing a form of cognitive flexibility that is particularly relevant in language. Humans excel at combining linguistic building blocks to infer the meanings of novel compositional words, such as “un-reject-able-ish”. The neural mechanisms and representations required for this ability remain unclear. To unravel these, we trained participants on a semi-artificial language in which the meanings of compositional words could be derived from known stems and unknown affixes, using abstract relational structure rules (e.g., “good-kla” which means “bad”, where “-kla” reverses the meaning of the stem word “good”). According to these rules, word meaning depended on the sequential relation between the stem and the affix (i.e., pre- vs. post-stem). During fMRI, participants performed a semantic priming task, with novel compositional words as either sequential order congruent (e.g., “short-kla”) or incongruent primes (e.g., “kla-short”), and real words serving as targets that were synonyms of the composed meaning of the congruent primes (e.g., “long”). Our results show that the compositional process engaged a broad temporoparietal network, while representations of composed word meaning were localized in a more circumscribed left-lateralized language network. Strikingly, newly composed meanings were decodable already at the time of the prime in a way that could not be accounted for representations of the prime words themselves. Finally, we found that the composition process recruited abstract rule representations in a bilateral frontoparietal network, in contrast to our preregistered prediction of a medial prefrontal-hippocampal network. These results support the hypothesis that people activate a bilateral frontoparietal circuitry for compositional inference and generalization in language.

Landau-Kleffner syndrome (LKS) is a rare epileptic syndrome causing language regression. In this preliminary study, we investigated the effects of simultaneous transcranial direct current stimulation (tDCS) and transcranial magnetic stimulation (rTMS) on LKS patients and the underlying mechanism based on magnetoencephalography (MEG) network analysis. Three LKS patients were included, all receiving daily stimulation for 15 days in 3 weeks, with MEG data collected to determine stimulation target and measure functional connectivity through normalized phase transfer entropy (PTE). The primary clinical outcome was measured through the Peabody Picture Vocabulary Test. Two patients with clustered spike dipoles in the right inferior parietal lobe (IPL) showed profound clinical improvement after stimulation. PTE analysis showed increased dorsal pathway information flow, mainly in the left hemisphere, and increased ventral pathway information flow in both hemispheres. This study preliminarily showed that tDCS-rTMS was safe to undergo and likely to improve language function in LKS patients. Stimulating the right IPL not only affects the right hemisphere language network but also profoundly affects the left hemisphere network that participates in language perception, processing, and understanding. The dynamic balance of the language network connections was reconstructed, which may be associated with improved language function in LKS. PLAIN LANGUAGE SUMMARY: Due to the lack of effective long-term treatment options, we investigated the effects of transcranial direct current stimulation (tDCS) and transcranial magnetic stimulation (rTMS) on language function in Landau-Kleffner syndrome (LKS) patients and the possible therapeutic mechanism by magnetoencephalography network analysis. Three LKS patients were included. We found tDCS-rTMS improved language function without side effects in two of the patients whose stimulation target was the right inferior parietal lobe, possibly due to significantly enhanced information flow in bilateral language streams, indicating modulated language network connections. Thus, we preliminarily proved the safety and effectiveness of tDCS-rTMS and unveiled underlying network mechanisms.

Product conceptual design is paramount in targeting specific customer segments, elevating brand equity, and ultimately boosting revenue sources. In the era of digitalisation, Generative Artificial Intelligence (GAI) is emerging to advance this process. This study introduces a pioneering framework, namely Generative Artificial Intelligence Product Designer (GAI-PD), that harnesses the power of GAI to revolutionise the design generation process, fostering efficiency, creativity, and responsible Artificial Intelligence (AI) adoption. GAI-PD is a novel multi-model framework integrating Generative Adversarial Network (GAN) and Large Language Model (LLM) that empowers designers to leverage GAN’s capabilities for realistic visual output and LLM’s natural language interaction. A comprehensive case example demonstrates the framework’s effectiveness, achieving 90.23% accuracy in generating valid design concepts and showing that hyperparameter tuning can significantly enhance image quality. From a managerial perspective, the findings indicate that adopting the GAI-PD framework can substantially reduce product development cycle times and accelerate time-to-market, offering a distinct competitive advantage. The study also explores avenues for future research, such as using advanced GAI models for high-resolution design generation and automated design evaluation, ultimately paving the way for enhanced design processes in diverse domains and promoting responsible AI adoption.

This work presents a computational framework for studying reaction dynamics via wavepacket propagation, employing the multiconfiguration time-dependent Hartree (MCTDH) method and its multilayer extension (ML-MCTDH) as the core methodologies. The core idea centers on the concept of modes that combine several coordinates along with their hierarchical separations because the degrees of freedom are too numerous to be efficiently treated as a single mode. First, the system is partitioned into several fragments within the same layer, and these fragments are further decomposed. Repeating this process, a hierarchical separation of modes emerges, until modes of a manageable size are achieved. Accordingly, the coordinate frame can be designed hierarchically. Second, the kinetic energy operator (KEO) is derived as a sum-of-products (SOP) of single-particle differential operators through a polyspherical approach, while the potential energy surface (PES) is expressed in a similar SOP form of single-particle potentials (SPPs) through (1) reconstruction approaches using an existing PES or (2) direct approaches based on a computed database. Third, the nuclear wave function is expressed in a multi-layer expansion form, where each term is a product of single-particle functions (SPFs) that are further expanded by the SPFs in the deeper layer. This expansion form is also adopted using a variational eigensolver for electronic wave function. Finally, the Dirac-Frenkel variational principle leads to a set of working equations whose solutions reproduce reaction dynamics, say reaction probability and time-dependent expectation. In addition, the hierarchical framework can be rearranged by the mathematical language of tensor networks (TN) or tree tensor networks (TTN). In this work, we compare the methods represented by a function with those in the form of a TN or a TTN. We also discuss the limitations of the present framework and propose solutions, providing further perspectives.

Altered effective connectivity in cortico-striatal pathways during sentence processing and oral motors in Parkinson's disease.

Electroconvulsive therapy improves functional connectivity of the frontoparietal network and language network in patients with treatment-resistant depression.

Background: Psychogenic Non-Epileptic Seizures (PNES), is a type of seizure that is caused by emotional factors. Symptoms of PNES are similar to epileptic seizures including disturbance in involuntary movement. Previous studies showed that neural activity altered in PNES detected through the resting-state functional Magnetic Resonance Imaging (rs-fMRI) thus this study was designed for a better understanding of PNES pathophysiology using the rs-fMRI technique. Objective: This study was conducted to examine dynamic Functional Connectivity (dFC) in the brain networks between PNES and healthy control subjects.Material and Methods: In this experimental study, the rs-fMRI was collected from 16 PNES subjects and 16 healthy subjects. After surrogating data, the sliding window technique was used for dFC detection in nine brain networks which chosen from Stanford Findlab. Results: Our results indicate that there were no differences in the presence or absence of dFC between the PNES group and the control group in the ventral Default Mode Network (vDMN), Language Network (LN), and Visuospatial Network (VSN). However, dFC was elevated in the PNES group in comparison to the normal control group within the Sensorimotor Network (SMN), Posterior Salience Network (PSN), and Anterior Salience Network (ASN). Conclusion: The findings suggest that dFC analyses hold significant potential for uncovering abnormal patterns of brain network connections in the PNES. This offers a promising finding for a better comprehension of PNES.