Biology subjects

Goldstein, A.

Publications and source records attributed to Goldstein, A..

At least 19 recordsLinked to original sources

bioRxiv2026

Beta bursts carry a graded signal of semantic updating during naturalistic narrative comprehension

Comprehending a narrative requires continuously updating a model of the discourse, and beta-band activity is thought to signal when that model should be held versus changed. Whether beta also grades how much to update, and whether that index is carried by discrete bursts beyond sustained power, is unknown for naturalistic comprehension. We recorded MEG from 61 human participants (22 men, 39 women) while they listened to naturalistic Hebrew narratives, detected beta bursts in the continuous signal, and quantified the semantic shift between successive sentences from sentence embeddings. Beta burst rate decreased after sentence boundaries, and regression-based deconvolution separated a pre-onset maintenance phase from a graded post-onset response whose magnitude scaled with semantic shift (peak ~0.9 s). This response was independent of lexical surprisal, the acoustic envelope, and referent introduction, largely robust to dependency structure, though it shared variance with propositional quantity. Burst occurrence carried a component of the effect beyond the accompanying amplitude change: applying that change as a pure gain to each participant's envelope and re-running the identical detection predicts only 75% of the observed burst-rate reduction (excess p = .038), the effect survived an amplitude-invariant threshold, and a Poisson model reproduced it as a graded reduction in burst-initiation rate. The effect held in both participant groups, which heard differently framed narratives, and was decisively supported (BF10 > 250). Beta bursts thus provide a graded signal of semantic updating during continuous comprehension, expressed in the rate of burst occurrence over and above the accompanying change in sustained power.

neuroscience↗

bioRxiv2026

Content-Sensitive Linguistic Representations in the Human Multiple-Demand Network

Neuroimaging dissociates specialized language regions from the domain-general multiple-demand (MD) network, yet the functional contribution of MD regions to language processing remains unresolved. Because MD recruitment during linguistic tasks is conventionally attributed to domain-general cognitive load, prior research has largely prioritized activation magnitude over representational content. Consequently, it remains unclear whether MD cortices function merely as nonspecific amplifiers of effort or actively encode fine-grained linguistic structures. Here we show that the MD network reliably encodes content-sensitive linguistic representations during naturalistic comprehension independently of peak cognitive demand. Using a voxelwise encoding framework with architecturally identical language models trained on distinct semantic domains (BERT and SciBERT) across three naturalistic fMRI datasets, we find that multiple-demand regions track linguistic structure during both high-demand scientific lectures and low-demand narratives. Furthermore, comparative analyses reveal that these representations are sensitive to domain-specific semantic regularities rather than surface-level form alone. Finally, by evaluating relative model alignment alongside the language network, we demonstrate that the representational balance between these two systems shifts dynamically as a function of task context and domain relevance, revealing a complementary division of labor. Together, these findings reframe the multiple-demand network from a purely extrinsic control mechanism into an active, content-sensitive component of a distributed semantic architecture.

neuroscience↗

bioRxiv2026

Modeling Genetic Diversity in Sickle Cell Disease Reveals Heterogeneous Responses to HbF-Inducing Therapies

Sickle cell disease (SCD) is caused by a point mutation in the {beta}-globin gene that promotes hemoglobin polymerization, leading to chronic hemolytic anemia, vaso-occlusive episodes, and progressive organ damage. The most efficacious therapies focus on reactivating fetal hemoglobin (HbF) expression to mitigate the pathological effects of sickle hemoglobin (HbS) polymerization. However, the predominantly used HbF inducer, hydroxyurea (HU), exhibits substantial interpatient variability in efficacy, and curative approaches such as gene therapy remain inaccessible to the vast majority of patients. Although all SCD patients share the same causative HBB glu7val mutation, differences in genetic background significantly influence disease severity and therapeutic response. We describe a SCD-specific induced pluripotent stem cell (iPSC) platform as a renewable and scalable preclinical model to interrogate treatment responses across the genetically diverse SCD patient population. By generating patient-specific iPSC-derived erythroblasts (iEry) representing distinct SCD genetic backgrounds, we demonstrate that this system faithfully recapitulates the heterogeneous HbF induction observed clinically in response to HU. Moreover, this platform enables the identification and evaluation of alternative therapeutic agents for HU non-responders and provides sufficient resolution to dissect drug-specific effects on erythroid differentiation and cellular phenotypes. Together, these findings support the use of iPSC-derived erythroid models as a versatile tool to advance precision therapeutic strategies for SCD. KEY POINTS- SCD iPSC-derived erythroid cells (iEry) reflect the diversity in HU-mediated HbF induction seen in SCD patients - SCD iEry recapitulate patient-specific treatment responses and can be used to identify therapeutic alternatives for HU non-responders - iEry provide a versatile platform to study the impact of novel HbF inducers on erythroid cell characteristics and differentiation parameters

cell biology↗

bioRxiv2025

The Road Not Taken: Unsaid Word Alternatives are Represented in the Brain

Human language allows multiple ways to express the same thought, implying that several lexical alternatives may exist in parallel before a single word is spoken or heard. We test the multiple alternatives-co-activation hypothesis by combining high-density ECoG during spontaneous dialogue with behavioral paradigms and ranked next-word predictions from large language models (LLMs). Behaviorally, words that LLMs rank as more likely continuations are recognized faster in a preregistered lexical decision task and produced with shorter inter-word intervals in free speech, indicating graded anticipatory activation of alternatives. Neurally, encoding models reveal that activity in classical language regions (IFG, STG) prior to word onset is predicted by embeddings of multiple top-ranked alternatives, not only by the word ultimately used; critically, mean embeddings that pool the top candidates outperform single-candidate embeddings, and the effect persists with arbitrary (non-semantic) embeddings that control for distributional similarity. Extending beyond a handful of options, encoding strength increases as embeddings are averaged across larger top-k sets, implying that a broad cohort of lexical candidates is simultaneously represented. Finally, models trained in comprehension generalize to production (and vice versa), preserving rank order and suggesting a shared neural code for candidate sets across modalities. Together, these findings provide direct evidence that the brain co-activates unsaid alternatives during natural language use and identify parallel candidate activation as a computational principle common to human comprehension, human production, and artificial language modeling.

neuroscience↗

bioRxiv2025

High-level Speech Processing During Mind-Wandering: Evidence from Neural Alignment with Language Models

Abundant evidence shows that when listening to speech or reading text, we continuously make predictions about upcoming words. Does this process stop when our mind wanders away and subjective experience becomes decoupled from the content of the speech? Conversely, is word prediction sufficient for conscious experience, or can these processes dissociate? Participants (N = 25) listened to audiobook content (> 12,000 words) while providing self-reports of mind-wandering. Mind-wandering was associated with spectral changes in the EEG signal, compared to attentive listening, as well as decreased early response to word onset, in line with current accounts. In contrast, EEG markers of word-level contextual surprise (indexed by the contextual surprise of large language models) remained intact during mind-wandering, alongside significant encoding of semantic content. Last, we measured the neural encoding of predictive context, measured by the correlation between EEG activity and the vectorial embeddings that language models use to predict each word given prior context. This encoding also persisted during mind-wandering, albeit weaker compared to attentive periods. Our findings show that the brain predicts and monitors relevant inputs even when the subjective experience of the external environment is dimmed during mind wandering. This suggests that the shared computational mechanisms between humans and Language Models are insufficient for having a subjective experience of the meaning of speech contents, with implications for theoretical accounts of mind-wandering and predictive processing accounts of consciousness. Significance StatementLarge Language Models (LLMs) achieve human-like conversational abilities by using long range context to predict the most likely next word. Such predictive processing is also common in the brain, and some link the best prediction to the current conscious experience. Using EEG, we show that next-word prediction markers, marked by LLM-brain alignment, persist during periods of mind-wandering while people listen to audiobooks. Specifically, we found that when people report mind-wandering, their brains continue to represent context-based predictions, word-level surprisal, and word-level semantics. These findings dissociate high-level predictions from subjective experience.

neuroscience↗

bioRxiv2025

Discovery of chromatin-based determinants of azacytidine and decitabine anti-cancer activity

The DNA-incorporating nucleoside analogs azacytidine (AZA) and decitabine (DEC) have clinical efficacy in blood cancers, yet the precise mechanism by which these agents kill cancer cells has remained unresolved - specifically, whether their anti-tumor activity arises from conventional DNA damage or DNA hypomethylation via DNA methyltransferase 1 (DNMT1) inhibition. This incomplete mechanistic understanding has limited their broader therapeutic application, particularly in solid tumors, where early clinical trials showed limited efficacy. Here, through the assessment of drug sensitivity in over 600 human cancer models and comparison to a non-DNA-damaging DNMT1 inhibitor (GSK-3685032), we establish DNA hypomethylation, rather than DNA damage, as the primary killing mechanism of AZA and DEC across diverse cancer types. In further support of an epigenetic killing mechanism, CRISPR drug modifier screens identified a core set of chromatin regulators, most notably the histone deubiquitinase USP48, as AZA and DEC protective factors. We show that USP48 is recruited to newly hypomethylated CpG islands and deubiquitinates non-canonical histones, establishing USP48 as a key molecular link between the two components of epigenetic gene regulation: DNA methylation and chromatin modification. Furthermore, loss of USP48, which occurs naturally through biallelic deletions in human cancers, sensitized both hematologic and solid tumors to DNMT1 inhibition in vitro and in vivo. Our findings elucidate the epigenetic mechanism of action of AZA and DEC and identify a homeostatic link between DNA methylation and chromatin state, revealing new therapeutic opportunities for DNMT1 inhibitors in solid tumors.

cancer biology↗

bioRxiv2025

Multivalent Lipid MVL5 Micellar Nanoparticles Exhibit Dramatically Increased Loading of Paclitaxel with PEGylation Enhancing Human Cancer Cell Penetration Depth and Cytotoxicity

Cationic liposomes (CLs) with chain-melted fluid membranes are promising nanocarriers of hydrophobic drugs in cancer chemotherapy, including the frequently employed drug paclitaxel (PTX). CL formulations containing univalent N-[2,3-dioleoyloxy-1-propyl]trimethylammonium chloride (DOTAP), like EndoTAG-1TM, have shown limited success in clinical trials and challenges like endosomal entrapment, limited PTX membrane solubility, and difficulty with in vivo tumor targeting remain. Incorporation of 10 mol% cone-shaped poly(ethylene glycol)-lipid (PEG-lipid) to DOTAP-containing CLs transitions a fraction of the particles to micellar nanodiscs. These PTX--loaded PEGylated CLs and nanodiscs show enhanced cellular uptake in vitro and improved tumor penetration and proapoptotic activity compared to bare CLs in an in vivo solid breast cancer tumor model. Formulations incorporating the multivalent cationic lipid MVL5 (+5e) at 50 mol% form nanoparticles (NPs) comprised almost entirely of nanodiscs, and transition at 75 mol% MVL5 to short micellar rods coexisting with spheres, with rods further transitioning to long flexible rods upon PEGylation. Here, we report on the finding that MVL5-based micellar NPs with disc, rod, and spherical morphologies dramatically improve the solubility of PTX in their fluid lipid membranes by nearly three-fold compared to reference CLs modeled on the EndoTAG-1TM formulation. Cell viability assays revealed that this improved PTX solubility for MVL5 micellar NPs leads to improved cytotoxic efficacy, which is further improved by PEGylation. Remarkably, using fluorescent microscopy and image particle analysis, we find that the cellular uptake and penetration depth of MVL5 nanoparticles is significantly improved by PEGylation. The findings are consistent with a model where the rate-limiting step of PTX delivery by cationic lipid NPs is diffusion of endocytic vesicles containing NPs through the actin mesh near the cell surface combined with the hoping rate of PTX from endosomal membrane to nearby microtubules. PEGylated cationic lipid nanoparticles containing MVL5 therefore represent a very promising hydrophobic cancer drug delivery vehicle for nanomedicine applications.

bioengineering↗

bioRxiv2025

A novel transgenic reporter to study vertebrate epigenetics

Epigenetic reprogramming contributes to the generation of cellular diversity during vertebrate development but the mechanisms directing this are still not well understood. Large-scale genetic screens have been highly successful in identifying epigenetic regulatory genes in invertebrates such as worms and flies, but similar large-scale genetic screens to identify epigenetic regulators have not been carried out in vertebrates. Here we report a newly generated "EpiTag" zebrafish transgenic reporter line that permits easy cellular-level visualization of epigenetic silencing or activation in living animals during development, gametogenesis, and regeneration. We use the EpiTag reporter to carry out an F3 ENU mutagenesis screen for epigenetic silencing or activating mutants, identifying relevant vertebrate tissue-specific epigenetic regulatory genes including a new epigenetic model for metabolic dysfunction-associated fatty liver disease (MAFLD). The EpiTag reporter line represents a powerful new tool for genetic and experimental analysis of tissue-specific epigenetic gene regulation in vertebrates. One Sentence SummaryEpiTag transgenic zebrafish provide a powerful new tool for visualizing and studying epigenetic regulation in living vertebrate animals.

genomics↗

bioRxiv2025

Early epigenetic priming of regeneration studied with a novel transgenic epigenetic reporter

Tissue regeneration requires previously differentiated cells to regain developmental plasticity. However, the upstream mechanisms initiating this process remain poorly understood. Here, we leverage a novel "EpiTag" transgenic zebrafish reporter line that enables real-time visualization of epigenetic silencing and activation to identify and carry out a comprehensive multi-omics analysis of cells undergoing epigenetic reprogramming during caudal fin regeneration. EpiTag GFP expression is transiently activated in cells contributing to regeneration between 12 to 16 hours post amputation (hpa), preceding the expression of canonical blastema markers. Single-cell RNA-seq reveals that GFP+ cells are restricted to regeneration-competent lineages such as pre-osteoblasts, proliferating cells, and wound epithelium. Integrated bulk RNA-seq, time-course RNA-seq, ATAC-seq, and bisulfite-seq on FACS-isolated GFP+ cells uncovers an early gene expression module enriched for chromatin regulators and a late gene expression module enriched for morphogenesis genes. Chromatin accessibility and DNA methylation changes are strongly associated with these late-expressed genes, suggesting epigenetic priming. We identify a number of epigenetic factors upregulated in the early gene expression module and show that ruvbl1 and ruvbl2, components of ATP-dependent chromatin remodeling complexes, are required for proper regeneration in both adult fins and larval tails. Our results establish EpiTag transgenics as a powerful in vivo tool for studying epigenetic reprogramming and highlight early chromatin remodeling events that enable activation of regenerative gene expression programs.

genomics↗

bioRxiv2025

Anatomical and Molecular Characterization of the Zebrafish Meninges

The meninges are a set of connective tissue layers that surround the central nervous system, protecting the brain from mechanical shock, supporting its buoyancy, guarding it from infection and injury, and maintaining brain homeostasis. Despite their critical role, the molecular identity, developmental origins, and functional properties of the cell types populating the meninges remain poorly characterized. This is in large part due to lack of cell type specific markers and difficulty in visualizing and studying these structures through the thick mammalian skull. Here, we show that the zebrafish, a genetically and experimentally accessible vertebrate, possesses an easily imaged mammalian-like meninges. Anatomical and cellular characterization of its composition via histology, electron microscopy, and confocal imaging shows that the adult zebrafish possesses complex multilayered meninges with double-layered dura mater and intricate leptomeningeal layers. Using single cell transcriptomics, we define the molecular identities of meningeal cell populations, including a unique ependymin (epd)-expressing cell population that constitutes the major cellular component of the leptomeningeal barrier and is essential for brain development and survival. These findings support the use of zebrafish as a useful comparative model for studying the meninges, provide a foundational description for future zebrafish meningeal research, and identify a new Leptomeningeal Barrier Cell that serves as the primary epithelial cell component of the leptomeninges.

developmental biology↗

bioRxiv2025

The "Podcast" ECoG dataset for modeling neural activity during natural language comprehension

Naturalistic electrocorticography (ECoG) data are a rare but essential resource for studying the brains linguistic capabilities. ECoG offers a high temporal resolution suitable for investigating processes at multiple temporal timescales and frequency bands. It also provides broad spatial coverage, often along critical language areas. Here, we share a dataset of nine ECoG participants with 1,330 electrodes listening to a 30-minute audio podcast. The richness of this naturalistic stimulus can be used for various research endeavors, from auditory perception to semantic integration. In addition to the neural data, we extract linguistic features of the stimulus ranging from phonetic information to large language model word embeddings. We use these linguistic features in encoding models that relate stimulus properties to neural activity. Finally, we provide detailed tutorials for preprocessing raw data, extracting stimulus features, and running encoding analyses that can serve as a pedagogical resource or a springboard for new research.

neuroscience↗

bioRxiv2024

Information-making processes in the speaker's brain drive human conversations forward

Natural human conversation is driven by the exchange of information-rich messages that surprise the listener and deviate from predictable context. While extensive research has characterized how the brain processes unexpected linguistic input during comprehension, the neural mechanisms underlying the generation of such information by the speaker remain poorly understood. In this study, we hypothesize that the speakers brain actively creates information through extra neural computation, transforming internal thoughts into novel, meaningful linguistic outputs. Utilizing a unique 24/7 dataset of continuous electrocorticography (ECoG) comprising approximately 100 hours of spontaneous natural conversations, we contrast the neural basis of speech production and comprehension within the same participants. Behaviorally, we find that speakers take longer pausing for an additional 100-150 milliseconds before producing information-rich, improbable words, even when controlling for word identity and frequency. Neurally, we provide converging evidence for a previously unreported process: generating information-rich words elicits significantly stronger neural activity (ERP) and enhanced neural encoding and decoding in language-related areas starting 100-500 ms before articulation. This pattern contrasts sharply with speech comprehension, where enhanced neural activity for predictable words occurs before onset, while responses to improbable words emerge only after onset as prediction errors. Furthermore, we demonstrate that large language models (LLMs) mirror this biological process, requiring deeper internal computation across layers to generate improbable versus probable words. Together, these results reveal that the speakers brain is not merely a transmission channel but an active generator of information, recruiting extra neural resources to create novel content that diverges from listener expectations.

neuroscience↗

bioRxiv2024

Scale matters: Large language models with billions (rather than millions) of parameters better match neural representations of natural language

Recent research has used large language models (LLMs) to study the neural basis of naturalistic language processing in the human brain. LLMs have rapidly grown in complexity, leading to improved language processing capabilities. However, neuroscience researchers havent kept up with the quick progress in LLM development. Here, we utilized several families of transformer-based LLMs to investigate the relationship between model size and their ability to capture linguistic information in the human brain. Crucially, a subset of LLMs were trained on a fixed training set, enabling us to dissociate model size from architecture and training set size. We used electrocorticography (ECoG) to measure neural activity in epilepsy patients while they listened to a 30-minute naturalistic audio story. We fit electrode-wise encoding models using contextual embeddings extracted from each hidden layer of the LLMs to predict word-level neural signals. In line with prior work, we found that larger LLMs better capture the structure of natural language and better predict neural activity. We also found a logarithmic relationship where the encoding performance peaks in relatively earlier layers as model size increases. We also observed variations in the best-performing layer across different brain regions, corresponding to an organized language processing hierarchy.

neuroscience↗

bioRxiv2024

ALIGNING BRAINS INTO A SHARED SPACE IMPROVES THEIR ALIGNMENT TO LARGE LANGUAGE MODELS

AO_SCPLOWBSTRACTC_SCPLOWRecent studies have shown that large language models (LLMs) can accurately predict neural activity measured using electrocorticography (ECoG) during natural language processing. To predict word-by-word neural activity, most prior work has estimated and evaluated encoding models within each electrode and subject--without evaluating how these models generalize across individual brains. In this paper, we analyze neural responses in 8 subjects while they listened to the same 30-minute podcast episode. We use a shared response model (SRM) to estimate a shared information space across subjects. We show that SRM significantly improves LLM-based encoding model performance. We also show that we can use this shared space to denoise the individual brain responses by projecting back into the individualized electrode space, and this process achieves a mean 38% improvement in encoding performance. The strongest improvement was observed for brain areas specialized for language comprehension, specifically in the superior temporal gyrus (STG) and inferior frontal gyrus (IFG). Critically, estimating a shared space allows us to construct encoding models that better generalize across individuals.

neuroscience↗

bioRxiv2024

YAP dysregulation triggers hypertrophy by CCN2 secretion and TGFβ uptake in human pluripotent stem cell-derived cardiomyocytes

Hypertrophy Cardiomyopathy (HCM) is the most prevalent hereditary cardiovascular disease - affecting >1:500 individuals. Advanced forms of HCM clinically present with hypercontractility, hypertrophy and fibrosis. Several single-point mutations in b-myosin heavy chain (MYH7) have been associated with HCM and increased contractility at the organ level. Different MYH7 mutations have resulted in increased, decreased, or unchanged force production at the molecular level. Yet, how these molecular kinetics link to cell and tissue pathogenesis remains unclear. The Hippo Pathway, specifically its effector molecule YAP, has been demonstrated to be reactivated in pathological hypertrophic growth. We hypothesized that changes in force production (intrinsically or extrinsically) directly alter the homeostatic mechano-signaling of the Hippo pathway through changes in stresses on the nucleus. Using human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs), we asked whether homeostatic mechanical signaling through the canonical growth regulator, YAP, is altered 1) by changes in the biomechanics of HCM mutant cardiomyocytes and 2) by alterations in the mechanical environment. We use genetically edited hiPSC-CM with point mutations in MYH7 associated with HCM, and their matched controls, combined with micropatterned traction force microscopy substrates to confirm the hypercontractile phenotype in MYH7 mutants. We next modulate contractility in healthy and disease hiPSC-CMs by treatment with positive and negative inotropic drugs and demonstrate a correlative relationship between contractility and YAP activity. We further demonstrate the activation of YAP in both HCM mutants and healthy hiPSC-CMs treated with contractility modulators is through enhanced nuclear deformation. We conclude that the overactivation of YAP, possibly initiated and driven by hypercontractility, correlates with excessive CCN2 secretion (connective tissue growth factor), enhancing cardiac fibroblast/myofibroblast transition and production of known hypertrophic signaling molecule TGF{beta}. Our study suggests YAP being an indirect player in the initiation of hypertrophic growth and fibrosis in HCM. Our results provide new insights into HCM progression and bring forth a testbed for therapeutic options in treating HCM. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=174 SRC="FIGDIR/small/597045v2_ufig1.gif" ALT="Figure 1"> View larger version (33K): org.highwire.dtl.DTLVardef@3357aforg.highwire.dtl.DTLVardef@c1c210org.highwire.dtl.DTLVardef@abb72forg.highwire.dtl.DTLVardef@187aa80_HPS_FORMAT_FIGEXP M_FIG C_FIG

cell biology↗

bioRxiv2024

Computation or Weight Adaptation? Rethinking the Role of Plasticity in Learning

The human brain is an adaptive learning system that can generalize to new tasks and unfamiliar environments. The traditional view is that such adaptive behavior requires a structural change of the learning system (e.g., via neural plasticity). In this work, we use artificial neural networks, specifically large language models (LLMs), to challenge this traditional view and suggest that such adaptive behavior can be also achieved solely through computation if the learning system is sufficiently trained. We focus on statistical learning paradigms. These require identifying underlying regularities in seemingly arbitrary word sequences and are primarily considered to require neural plasticity. LLMs can capture arbitrary structures without requiring weight adaptation, despite divergence from their natural language training data. Our work provides novel insights into the role of plasticity in learning, demonstrating that sufficiently trained learning systems are highly flexible, adapting to new tasks and environments solely through computation, much more than previously acknowledged. Furthermore, our work opens the door for future research to use deep learning models to conjure hypotheses about the brain.1

neuroscience↗

bioRxiv2024

A call for a unified and multimodal definition of cellular identity in the enteric nervous system

The enteric nervous system (ENS) is a tantalizing frontier in neuroscience. With the recent emergence of single cell transcriptomic technologies, this rare and poorly understood tissue has begun to be better characterized in recent years. A precise functional mapping of enteric neuron diversity is critical for understanding ENS biology and enteric neuropathies. Nonetheless, this pursuit has faced considerable technical challenges. By leveraging different methods to compare available primary mouse and human ENS datasets, we underscore the urgent need for careful identity annotation, achieved through the harmonization and advancements of wet lab and computational techniques. We took different approaches including differential gene expression, module scoring, co-expression and correlation analysis, unbiased biological function hierarchical clustering, data integration and label transfer to compare and contrast functional annotations of several independently reported ENS datasets. These analyses highlight substantial discrepancies stemming from an overreliance on transcriptomics data without adequate validation in tissues. To achieve a comprehensive understanding of enteric neuron identity and their functional context, it is imperative to expand tissue sources and incorporate innovative technologies such as multiplexed imaging, electrophysiology, spatial transcriptomics, as well as comprehensive profiling of epigenome, proteome, and metabolome. Harnessing human pluripotent stem cell (hPSC) models provides unique opportunities for delineating lineage trees of the human ENS, and offers unparalleled advantages, including their scalability and compatibility with genetic manipulation and unbiased screens. We encourage a paradigm shift in our comprehension of cellular complexity and function in the ENS by calling for large-scale collaborative efforts and research investments.

neuroscience↗

bioRxiv2024

An Incremental Large Language Model for long text processing in the Brain

Accumulated evidence suggests that Large Language Models (LLMs) are beneficial in predicting neural signals related to narrative processing. The way LLMs integrate context over large timescales, however, is fundamentally different from the way the brain does it. In this study, we show that unlike LLMs that apply parallel processing of large contextual windows, the incoming context to the brain is limited to short windows of a few tens of words. We hypothesize that whereas lower-level brain areas process short contextual windows, higher-order areas in the default-mode network (DMN) engage in an online incremental mechanism where the incoming short context is summarized and integrated with information accumulated across long timescales. Consequently, we introduce a novel LLM that instead of processing the entire context at once, it incrementally generates a concise summary of previous information. As predicted, we found that neural activities at the DMN were better predicted by the incremental model, and conversely, lower-level areas were better predicted with short-context-window LLM.

neuroscience↗