Neurons and Cognition · Cross-listed from cs.SI
Higher-order rich clubs and configuration models on general directed hypergraphs
Jason P. Smith, Celia Hacker, Jānis Lazovskis et al. (+3 more)
Detecting structure in complex networks, especially those arising from physical systems, is a central problem across the sciences. One approach is via rich club analysis, which identifies important vertices using a centrality metric and measures whether those vertices are more tightly interconnected than expected by chance. While informative, this approach captures only pairwise interactions, missing out on higher-order ones known to shape the structure and function of many complex systems. We propose a hyper-rich club pipeline that asks whether central vertices are more tightly interconnected than expected by chance through hyperedges encoding higher-order interactions, which also enables the inclusion of important, often omitted, directional information. We work in a broad class of hypergraphs, which we call general directed hypergraphs, that includes as special cases undirected hypergraphs, head-and-tail directed hypergraphs, and totally ordered hypergraphs (a hypergraph related to directed simplicial complexes from topological data analysis). This unifies several non-equivalent notions of directed hypergraph under one definition. On these hypergraphs we define a hyper-rich club framework whose concrete construction depends on explicit choices the domain scientist fixes according to their research goals. Particular choices recover the existing rich club notions for graphs and undirected hypergraphs, and yield the first such notion for each version of directed hypergraphs. We demonstrate that the pipeline recovers meaningful structure in data by studying networks of very different origins: connectomes, temporal networks of infectious spread, networks of poems, and the XGI hypergraph database, in each case detecting structure the standard graph rich club misses.
Read the full abstract ▾
All authors Jason P. Smith, Celia Hacker, Jānis Lazovskis, Florian Unger, Keith M. Smith, Daniela Egas Santander
Genomics
CLARA: Clarification of Language Ambiguity through Result Analysis for Natural-Language Cancer Genomics Queries
Pratyush Kumar Shukla, Manveer Singh Tib and Siddhant Garg
A natural
language interface can be used to make cancer genomics databases easier to use, but even if a question is perfectly fluent, its scientific meaning can be ambiguous. We propose CLARA, a framework that represents a question as a typed scientific query specification, considers a few possible interpretations, executes them, and asks for clarification when the estimates diverge. CLARA was assessed on mutation-prevalence contrasts among eight TCGA PanCancer Atlas cohorts and a 30-gene panel. This
benchmark consisted of 330 unique executable contrasts varying in mutation scope, assay denominator, and sample context; 115 contrasts were result-sensitive and 215 were result-stable, per the preregistered definition of relative divergence greater than 0.10 or absolute divergence greater than 5 percentage points. An independently implemented pandas execution engine perfectly replicated all 660 results from the SQLite engine. In a separate 120-question LLM-generated, manually vetted
language stress test, CLARA recognized all 60 result-sensitive contrasts and needlessly clarified 13 of 60 stable contrasts (accuracy 89.2%, sensitivity/recall 100%, specificity 78.3%). Standalone machine learning had superior overall accuracy (97.5%) but missed one critical contrast. This demonstrates that downstream execution can distinguish consequential from inconsequential ambiguity and reveal an explicit trade-off between safety and burden.
Read the full abstract ▾
Genomics · Cross-listed from cs.LG
Scaling an Autoregressive Transformer for Single-Cell Generation
Aleksandr Sharipov, Yusif Mukhtarov and Igor Molybog
We study a self-supervised generation task for single-
cell gene expression vectors: given a set of vectors from a
cell type, we aim to generate additional gene expression vectors of that
cell type. For this task we characterize both the biological fidelity of the generated gene expression vectors and the scaling behavior of the pretraining loss. The model is a causal
transformer paired with a learned quantized VAE tokenizer, trained with a cross-
entropy loss. To evaluate the model, we condition it on held-out gene expression vectors of a
cell type and generate vectors of gene expression, comparing the resulting distribution over gene expression vectors to the ground truth distribution of that
cell type. We study the scaling properties of the proposed architecture by varying the number of trained parameters and the amount of training data. To our knowledge, we find the first jointly-fit two-exponent scaling law and compute-optimal frontier for a single-
cell foundation model. Finally, we discuss how this pretrained model could be finetuned for perturbation response prediction.
Read the full abstract ▾
Neurons and Cognition · Cross-listed from stat.AP
Modelling temporal dynamics of suicidal ideation and behaviour across pre- to early adolescence using a Markov framework
Sieun Lee, Ben Cardoen, Marianne Etherson et al. (+10 more)
Understanding the dynamics of suicidal ideation and behaviour in youth and the factors associated with transitions from thoughts to behaviours is critical for early identification, monitoring, and prevention. Using longitudinal self-report data from the Adolescent
Brain Cognitive Development (ABCD) Study (n = 11,864) spanning ages 9 to 13 years, we developed a time-inhomogeneous discrete-time Markov chain framework to model transitions across eight states defined by suicidal ideation and behaviour and the co-reported presence or absence of non-suicidal self-injury (NSSI). The framework enables inference of year-to-year and multi-year transition probabilities, quantification of transition stability and uncertainty, and statistical comparison of transition likelihoods. We identified structured but developmentally changing transition patterns, including a generally high probability of recovery to a no-report state following reports of suicidal ideation or behaviour. Co-reported lifetime NSSI marked a distinct trajectory profile characterized by both greater risk and lower predictability: children with NSSI were more likely to transition to or persist in suicidal behaviour, were less likely to recover to a no-report state, and exhibited greater uncertainty in their transition likelihoods compared to those reporting suicidal ideation and/or behaviour alone. These findings suggest that NSSI marks not only elevated risk but also greater trajectory instability during pre- to early adolescence. By estimating probabilities of escalation, remission, persistence, and fluctuation in longitudinal cohort data, the framework provides a systematic, interpretable, and scalable approach to characterising suicidal trajectory dynamics in large, sparse mental health
datasets, with potential to inform future research on monitoring, early prevention, and time-varying risk in youth mental health.
Read the full abstract ▾
All authors Sieun Lee, Ben Cardoen, Marianne Etherson, Nitish Jawahar, Ellen Townsend, Kapil Sayal, Peter Fonagy, Aja Murray, Joanna Lockwood, Ayan Mahamud, Chris Hollis, Rory O'Connor, Dorothee Auer
Biomolecules
Expanding Protein Structure Prediction into Conformational State Space
Devlina Chakravarty, Justin J. Miller, Da Teng et al. (+8 more)
Recent AI advances have enabled
protein structure prediction at near-experimental accuracy, largely solving the problem of identifying a dominant conformation from sequence. Many
proteins, however, function as dynamic systems populating multiple conformational states with activity emerging from shifts in relative occupancy–an incomplete picture when reduced to one structure. Here, we argue that structure prediction should be reformulated as a state-space inference problem: recovering not one conformation's coordinates but accessible states, their energetic and kinetic relationships, context dependence, and responses to perturbations. We review emerging strategies–deep learning ensemble generators, physics-based simulations, and experimental constraints–and outline a roadmap toward state-space prediction.
Read the full abstract ▾
All authors Devlina Chakravarty, Justin J. Miller, Da Teng, Yousuf O. Ramahi, Patrick Bryant, Camila Neira-Mahuzier, César A. Ramírez-Sarmiento, Sarah Rauscher, Gregory R. Bowman, Pratyush Tiwary, Lauren L. Porter
Biomolecules · Cross-listed from physics.chem-ph
A concentration-independent paradigm rendering weak interactions inherently quantifiable
Masahiko Yoshimura, Fuyuki Matsuda, Yoshiki Ikeda et al. (+9 more)
A vast class of weak, millimolar-affinity molecular interactions governs cellular function, yet their quantitative characterization has remained largely beyond conventional methods. For over a century, biochemistry has worked within a concentration-based framework where molarity scales with molecular number per volume (N/V), and experiments have usually, often implicitly, changed concentration by moving N while holding V fixed. The weak-interaction measurement bottleneck arises from this paradigm: reading weak binding through bulk concentration requires concentrations beyond practical limits, a framework constraint rather than one of instrumental sensitivity. Here we show that shifting experimental control from N to accessible volume V overcomes this bottleneck and opens previously intractable affinity ranges through nanoscale spatial confinement. Controlling V means controlling what biochemists have called "local concentration" and "proximity effects," recasting these long-ambiguous notions as quantitative variables grounded in first principles. Implemented in DNA nanocavities, the approach showed that
geometric arrangement alone can override solution-phase binding hierarchies. The same spatial control quantified a
protein-peptide interaction of order 10 mM from femtomoles per well, totalling under a picomole per titration. Even so, a standard plate reader gave a signal-to-noise ratio near 10^3, leaving headroom for still weaker interactions. The affinity-and-geometry readout also enabled rational screening for
protein-
protein-interaction modulators, identifying compounds that enhance weak associations by reweighting local encounters rather than binding tightly on their own or forming a stable ternary complex. Together, this volume-based paradigm and its implementation provide a general strategy for probing and modulating previously inaccessible biochemical phenomena.
Read the full abstract ▾
All authors Masahiko Yoshimura, Fuyuki Matsuda, Yoshiki Ikeda, Chihiro Mori, Tomoko Yoneda, Minako Kikukawa, Rie Murakami, Chiharu Nogami, Yukihiko Sugita, Yoshiko Nakada-Nakura, Masahiko Tsujimoto, Daishi Fujita
Biomolecules · Cross-listed from physics.chem-ph
Experimental access to molarity's blind spot in macroscopic assays
Fuyuki Matsuda, Masahiko Yoshimura, Shiro Ikeda et al. (+1 more)
Chemical kinetics has long inferred local molecular behaviour through the flask-and-molarity pairing, where well-mixed concentrations serve as the experimental readout. Yet many biological reactions occur in structured environments. Researchers have long recognized that concentration may not carry the same operational meaning in such environments, but even local concepts such as effective molarity usually translate local effects back into a single value with units of concentration. What has been missing is the complementary path: a bench-compatible way to make local structure an experimental variable, rather than only a correction to molarity. Here we show a chemistry-geometry crossover that the flask-and-molarity interface could not make visible. In the micromolar-or-weaker affinity regime, inhibition can switch sharply out of the familiar concentration-and-affinity mode: chemical binding strength no longer determines the response, and the shape of the target's local space does. A bench-compatible interface made this switch measurable by separating bulk dose from local geometry. This blind spot arose from the hidden premise that macroscopic pooling makes a structured local state readable as a single local concentration. The chemistry-geometry crossover breaks that premise: in a structured target environment, a macroscopic assay can remain sensitive to the probability distribution of local states, so collapsing that distribution to one concentration-valued number removes the
geometric control axis from the readout. By preserving that axis in the experiment, the interface bypasses molarity's hidden bottleneck and provides a routine experimental route to remeasure and reinterpret molecular interactions in structured space.
Read the full abstract ▾
All authors Fuyuki Matsuda, Masahiko Yoshimura, Shiro Ikeda, Daishi Fujita
Neurons and Cognition
Detecting high-frequency brain disorder signals using dynamic mode decomposition from EEG
Jacob Kang and Jong-Hyeon Seo
Recent studies have reported clearly identifiable dynamical changes in the high-frequency range of
EEG signals recorded during specific stimuli, such as visual or auditory inputs, or in cases of
brain disorders like epileptic seizures. In this study, we utilized Dynamic Mode Decomposition (DMD) to extract consistent and persistent dynamical changes in the high-frequency band from the signals of neurologically relevant
EEG channels. High-frequency DMD modes were employed as features, composing a feature table. Through post-processing, a random distribution test was performed, revealing that approximately 70% of the samples exhibited consistent high-frequency dynamics within the signal of a specific channel. Furthermore, classification experiments confirmed that the PCA components of the feature table that passed the test formed a consistent pattern that distinguished the alcohol-dependent group from the control group.
Read the full abstract ▾
Quantitative Methods
REDE: A Quantitative Framework for Differential-Expression Reproducibility and Diagnostic Transfer Across Nine Cohorts and Three Cancers
Athanasios Angelakis
Differential-expression analyses often turn cohort-specific significance into claims of stable gene signatures or diagnostic biomarkers. We evaluated which layers of evidence reproduce across independent
datasets and whether discovery-derived panels retain locked tumor-versus-non-tumor classification performance. Nine public microarray cohorts were organized into fixed discovery, validation, and external-test experiments for pancreatic ductal adenocarcinoma, breast cancer, and lung cancer. Reproducibility was assessed for DEG burden, exact membership, top-rank overlap, signed effects, prespecified gene confirmation, and Hallmark pathways. We also introduced REDE-2Fold, in which each discovery cohort is split once at patient level, differential expression is performed independently in both folds, and only same-direction genes selected in both are retained. Four training-only panels were then evaluated with locked logistic models and thresholds: all discovery DEGs, the top 19 discovery DEGs, all REDE-2Fold genes, and the top 19 REDE-2Fold genes. Broad DEG-list confirmation in both independent cohorts ranged from 15.5% to 39.5%, rising to 50.1% to 84.3% for large effects. Pathway replication ranged from 52.2% to 88.9%. Compact 19-gene panels retained high external ROC-AUC, but locked operating points were often unstable: some models with ROC-AUC near 1.0 showed zero specificity or very low sensitivity. These results define a hierarchy from thresholded membership through effect, pathway, discrimination, and operating-point transfer. REDE provides a seven-level quantitative framework for matching transcriptomic claims to the evidence actually tested, while REDE-2Fold offers a minimum internal feature-stability procedure that strengthens but does not replace independent validation.
Read the full abstract ▾
Quantitative Methods
A method for comparing inferred evolutionary accumulation dynamics across covariates and model structures
Jakob Fernø, Michelle B. Verstraaten, Lila Gravellier et al. (+3 more)
In this note we describe a method for comparing inferred dynamics in evolutionary accumulation models (EvAMs). These models involve the acquisition of multiple, potentially codependent, binary features over time – for example, mutations in cancer development, or phenotypes in evolutionary biology. As the set of methods for inferring EvAM dynamics expands, approaches for comparing inference across
algorithms,
datasets, and
covariates are required. In particular, a comparison method supporting reversible,
stochastic dynamics, interactions between feature sets, and potentially non-independent samples (as well as simpler cases) has yet to be established. It is possible for EvAM with similar relative feature orderings to produce completely different sets of observed states due to “frameshift”-like differences; methods distinguishing state and transition similarity are therefore also desirable. Here we suggest a method focussed on ordering
matrices, describing the probability that a feature is acquired under different conditions on other features, and is thus generally comparable across methods and
datasets. We demonstrate how the approach captures both statistical robustness (given EvAM uncertainty) and scientifically meaningful differences in inferred dynamics. The method is applied to synthetic and real-world data on the
evolution of chromosomal aberrations in different tumour types and of drug-resistant bacteria in different countries.
Read the full abstract ▾
All authors Jakob Fernø, Michelle B. Verstraaten, Lila Gravellier, Ellen C. Røyrvik, Ramon Diaz-Uriarte, Iain G. Johnston
Neurons and Cognition · Cross-listed from cs.AI
Predictive Set Theory: A Generative Framework for Cognitive Architecture with Operationalized Core Mechanisms
Yiyang Yu
Predictive processing theories portray the
brain as a hierarchical prediction engine that minimizes prediction error, yet they lack operational definitions for the structure of a "prediction," the standardized response to a prediction error, and the mechanism that maintains consistency across successive updates.
Bayesian cognitive science attempts to subsume all uncertainty under probabilistic belief updating, but it presupposes a closed hypothesis space and provides no generative account of how the objects over which probabilities are distributed become discrete, identifiable referents in the first place. This paper introduces Predictive Set Theory (PST), a formal generative framework that reconstructs cognitive architecture from first principles. PST anchors cognition in a minimal set of operations—a sensor formalized as an identity function, set-theoretic state refresh, and three fundamental forms of reference chains (reference, counter-reference, and semi-reference)—and rigorously derives core cognitive functions including state sequences, demand, comparison, efficiency, and finite-horizon probabilistic planning. Rather than modeling neural mechanisms, PST constitutes a design specification for any system that must maintain internal consistency while acting under incomplete information and irreversible risk. The framework offers novel resolutions to classical problems such as Russell's paradox, the cognitive status of Gödelian incompleteness, the grounding of negative feedback, and the comprehension of film editing. The primary purpose of this paper is to establish, through the public academic record, the originality and completeness of the Predictive Set Theory framework.
Read the full abstract ▾
Quantitative Methods
A Blind Spot in Alignment: Quantifying Biosecurity Risks in Large Language Models
Shu Quan, Tianfang Hao, Sitong Fang et al. (+8 more)
Large
Language Models (LLMs) are accelerating biological research, yet this same capability poses a critical biosecurity threat: models that assist in
protein engineering can equally be prompted to generate predicted toxin-like sequences, potentially lowering the barrier to biological misuse. Current safety evaluations, however, operate in natural
language and cannot determine whether a model-generated amino acid sequence is biological gibberish or a computational risk signal. To address this evaluation blind spot, we introduce SPIKE-Bench, coupling 631 curated toxin-design
prompts across seven functional categories with the SPIKE funnel, a three-stage protocol that filters output through compliance, biological plausibility, and predicted toxicity, producing stage-level diagnostics and an aggregate function-aware metric: the Functional Harmfulness Rate (FHR). An audit of 32 LLMs reveals that most models freely comply with toxin-design requests; FHR is driven primarily by biological generation capability rather than safety
alignment, reaching 50.7%; and Refusal Rate fails to predict functional risk. As a first step toward mitigation, we provide BioSafe-Guard, a domain-specialized classifier that substantially reduces predicted functional risk while preserving benign utility. We release SPIKE-Bench and BioSafe-Guard at https://github.com/PKU-
Alignment/SPIKE-Bench to support more rigorous biosecurity evaluation of LLMs.
Read the full abstract ▾
All authors Shu Quan, Tianfang Hao, Sitong Fang, He Geng, Jiayi Zhou, Boyuan Chen, Kaile Wang, Donghai Hong, Juntao Dai, Yaodong Yang, Jiaming Ji
Quantitative Methods
A curvature-based criterion for harmonic circadian waveforms
Yusuke Yamada, Yutaro Kabata, Ryoya Fukasaku et al. (+1 more)
Experimental and theoretical studies of circadian rhythms have focused largely on the period, on mutants that alter it and on phase shifts, and this focus has driven the identification of clock genes and clarified how clocks entrain to light–dark cycles. The waveform of the oscillation itself, by contrast, has attracted little
attention as an indicator of the properties of the underlying oscillator. To assess the waveform directly, we focus on whether the trajectory in the plane spanned by a variable and its time derivative possesses an inflection point, and we define an oscillation to be harmonic when no inflection point is present. Bioluminescence recordings from cyanobacteria and from the mammalian SCN were harmonic in this sense, as were most of the core clock components in mathematical models of the circadian clock. Numerical analysis of the Goodwin model, a minimal representation of the core circadian oscillator, yielded harmonic oscillation throughout. We confirmed this numerical trend semi-analytically using a piecewise-linearized Goodwin model. Because it evaluates the properties of a waveform without assuming a model structure, the approach we propose offers a new perspective on the waveform analysis of biological rhythms in general, not only circadian ones.
Read the full abstract ▾
All authors Yusuke Yamada, Yutaro Kabata, Ryoya Fukasaku, Hiroshi Ito
Quantitative Methods
NeuroInspector: A Local-First Environment for Inspecting and Annotating Hierarchical Neuroscience Datasets
Zihan Yang
The growing scale and structural complexity of neuroscience
datasets have made
dataset inspection an increasingly distinct stage of the research workflow. Existing inspection workflows, however, remain fragmented, often relying on exploratory scripts, manual documentation, and repeated navigation of unfamiliar file structures before meaningful scientific analysis can begin. Here we present NeuroInspector, a lightweight, browser-based environment for inspecting HDF5 and NWB files. Using WebAssembly-based HDF5 parsing (h5wasm), NeuroInspector runs entirely client-side: files are opened directly from local disk, and the application implements no file-upload endpoint, including in its hosted deployment. The tool combines structural navigation, metadata inspection, sampled data previews, and path-level annotation into portable, fingerprinted "project packs" that preserve inspection decisions without modifying the original file. Rather than functioning as an analysis or validation platform, NeuroInspector provides a dedicated, traceable environment for the inspection stage that precedes formal analysis.
Read the full abstract ▾
Quantitative Methods · Cross-listed from cs.CV
Self-supervised DXA representations encode multi-system disease risk, biological aging and heritability
Gil Sasson, Zachary Levine, Smadar Shilo et al. (+9 more)
Whole-body dual-energy X-ray absorptiometry (DXA) scans are routinely acquired to measure bone density and regional body composition, leaving their spatial structure largely unused. Here, we show that self-supervised learning (SSL) can convert raw DXA images into representations of systemic health. We introduce LeDXA, a
vision model based on a joint-
embedding predictive architecture (JEPA) that learns by predicting latent representations rather than reconstructing pixels. Trained from scratch on 11,540 unlabeled Human Phenotype Project scans, LeDXA was evaluated internally and on 47,400 external UK Biobank (UKBB) scans. It improved cross-cohort prediction of prevalent diseases and biomarkers beyond scanner-derived DXA measurements and DINOv3, a state-of-the-art general-purpose model, despite approximately 150,000-fold fewer training images and nearly 40-fold fewer parameters. Over a median 4.3-year UKBB follow-up, LeDXA improved incident disease prediction over tabular DXA measures, with the largest gains for hip and knee arthrosis and type 2 diabetes. For hip arthrosis, 66% of incident cases occurred in the highest-risk quartile versus 41% for tabular measures. Its representations predicted chronological age externally (r = 0.88; mean absolute error = 2.90 years), and the biological-age gap tracked broader disease burden and a 45% higher mortality hazard in the oldest-appearing quartile. The gap also decreased in women after starting hormone-replacement therapy, suggesting it may be modifiable. Genome-wide associations recovered mostly known body-composition and bone-density loci, and LeDXA
embeddings were more heritable than DINOv3's. These findings reveal prognostic information in DXA images that conventional readouts discard, learnable with relatively little data and modest compute.
Read the full abstract ▾
All authors Gil Sasson, Zachary Levine, Smadar Shilo, Sarah Kohn, Guy Lutsker, Anastasia Godneva, Adam Gabet, David Krongauz, Adina Weinberger, Yann LeCun, Randall Balestriero, Eran Segal
Quantitative Methods
Phylogeny.fr: the phylogenetic platform designed for non-specialists
Valentin Gorgodian, Olivier Poirot, Alain Schmitt et al. (+5 more)
Phylogenetic analysis has become a standard approach across many areas of biology, yet the growing complexity of phylogenetic methods and software remains a major obstacle for non-specialists. Since its launch in 2008, Phylogeny.fr has provided an accessible web platform for building phylogenetic trees using widely accepted methods without requiring local software installation. Here, we present a major redesign and modernization of the service. The new version integrates state-of-the-art tools while preserving historical programs for legacy support and relies on modern web architecture and HPC infrastructure. New interactive React-based viewers, ReSeqt and Reactree, provide intuitive exploration and publication-ready visualization of
alignments and trees. The Blast-Explorer companion tool has also been updated and now includes clustering options. By combining ease of use, methodological flexibility, and modern phylogenetic tools, the new Phylogeny.fr addresses the needs of researchers, teachers, and students seeking accessible and reliable phylogenetic analyses.
Read the full abstract ▾
All authors Valentin Gorgodian, Olivier Poirot, Alain Schmitt, Virginie Collomb, Jean-Michel Claverie, Chantal Abergel, Matthieu Legendre, Sebastien Santini
Neurons and Cognition
Divisive Normalization Shapes Low-Rank Slow Manifolds for Continuous Working Memory
Zhaotian Gu, Jie Su, Weiwei Wang et al. (+3 more)
The ability to robustly maintain and update continuous variables is a hallmark of working
memory. While classical continuous attractor networks suffer from severe fine-tuning fragility, standard artificial recurrent neural networks (RNNs) like GRUs and LSTMs typically fail to stably learn continuous
manifolds, instead shattering the state space into discretized point attractors. To bridge this gap, we draw inspiration from divisive normalization, a canonical neural computation widely observed across cortical
circuits, and propose the Recurrent Divisive Normalization Network (RDNN), a minimal and algebraically isolated model of dynamic division. Through dynamical systems analysis on canonical working
memory tasks, we demonstrate that this biophysical constraint allows the network to converge to robust, high-fidelity slow
manifolds. Furthermore, we analyze the gradient dynamics of divisive normalization during Backpropagation Through Time (BPTT), showing that it introduces an activity-dependent local gradient scaling. This scaling dampens parameter updates in highly active regimes, which empirically aligns with a significant self-compression of the network's effective rank, confining the recurrent dynamics to a tight, low-dimensional subspace while avoiding the optimization pathologies associated with explicit low-rank factorization. Finally, ablations demonstrate that while subtractive inhibition can maintain static
memories, divisive normalization is mathematically essential to prevent
manifold shattering under time-varying inputs. Our findings identify divisive normalization not merely as a biological artifact, but as a critical computational mechanism for learning high-fidelity continuous representations.
Read the full abstract ▾
All authors Zhaotian Gu, Jie Su, Weiwei Wang, Chang Liu, Tianyi Qian, Dahui Wang
Neurons and Cognition
NeuroWorld: A Latent Brain World Model for Stimulus-Conditioned Human Brain Dynamics
Zijian Dong, Jianxiong Zhou, Kwun Kei Ng et al. (+4 more)
Forecasting human
brain activity during naturalistic experience requires modeling how endogenous neural states evolve causally under continuous sensory drive. Existing
brain encoding models instead frame this as stimulus-to-response regression without strict temporal constraints, allowing future stimuli to leak into current predictions. We introduce NeuroWorld, to our knowledge the first
brain world model, which casts naturalistic
brain functional dynamics prediction as stimulus-conditioned
evolution in a learned latent
brain-state space, separating endogenous states (measured via fMRI) from exogenous
multimodal stimuli across two stages. Latent Dynamics Learning (LDL) jointly learns a transition-sufficient representation and causal dynamics through next-latent prediction, without reconstructing the observed fMRI signal. Latent Rollout Decoding (LRD) freezes LDL, autoregressively rolls latent states forward from an observed fMRI prefix, and decodes them into subject-specific whole-
brain responses. Across three naturalistic movie-fMRI
benchmarks spanning 30 participants, including our newly collected Singapore
Multimodal Imaging & Naturalistic
Dataset (SG-MIND; 20 participants, 8,519 paired stimulus-response clips, 140.7 person-hours of viewing), NeuroWorld achieves state-of-the-art multi-step rollout performance under strictly causal stimulus access, with greater robustness to long-horizon autoregressive drift, supporting reliable simulation of extended
brain-state trajectories. Extensive interpretability analyses characterize the functional organization of the learned dynamics, establishing latent-space world modeling as a principled framework for causal forecasting of human
brain activity.
Read the full abstract ▾
All authors Zijian Dong, Jianxiong Zhou, Kwun Kei Ng, Jan Paolo Macapinlac Balagtas, Zhizhou Li, Zijiao Chen, Juan Helen Zhou