Rapid adaptation in complex control systems remains a central challenge in reinforcement learning. We introduce a framework in which policy and value functions share a low-dimensional coefficient vector - a goal embedding - that captures task identity and enables immediate adaptation to novel tasks without retraining representations. During pretraining, we jointly learn structured value bases and compatible policy bases through a bilinear actor-critic decomposition. The critic factorizes as Q = sum_k G_k(g) y_k(s,a), where G_k(g) is a goal-conditioned coefficient vector and y_k(s,a) are learned value basis functions. This multiplicative gating - where a context signal scales a set of state-dependent bases - is reminiscent of gain modulation observed in Layer 5 pyramidal neurons, where top-down inputs modulate the gain of sensory-driven responses without altering their tuning. Building on Successor Features, we extend the decomposition to the actor, which composes a set of primitive policies weighted by the same coefficients G_k(g). At test time the bases are frozen and G_k(g) is estimated zero-shot via a single forward pass, enabling immediate adaptation to novel tasks without any gradient update. We train a Soft Actor-Critic agent on the MuJoCo Ant environment under a multi-directional locomotion objective, requiring the agent to walk in eight directions specified as continuous goal vectors. The bilinear structure allows each policy head to specialize to a subset of directions, while the shared coefficient layer generalizes across them, accommodating novel directions by interpolating in goal embedding space. Our results suggest that shared low-dimensional goal embeddings offer a general mechanism for rapid, structured adaptation in high-dimensional control, and highlight a potentially biologically plausible principle for efficient transfer in complex reinforcement learning systems.

Normalization is a critical operation in neural circuits. In the brain, there is evidence that normalization is implemented via inhibitory interneurons and allows neural populations to adjust to changes in the distribution of their inputs. In artificial neural networks (ANNs), normalization is used to improve learning in tasks that involve complex input distributions. However, it is unclear whether inhibition-mediated normalization in biological neural circuits also improves learning. Here, we explore this possibility using ANNs with separate excitatory and inhibitory populations trained on an image recognition task with variable luminosity. We find that inhibition-mediated normalization does not improve learning if normalization is applied only during inference. However, when this normalization is extended to include back-propagated errors, performance improves significantly. These results suggest that if inhibition-mediated normalization improves learning in the brain, it additionally requires the normalization of learning signals.

Understanding infectious disease transmission in institutional settings requires models that capture how contacts arise from structured routines, roles, and spatial constraints. In aged care facilities, interactions are driven by care delivery, staff scheduling, and resident mobility, producing patterns that differ from those assumed in population-level models. This study develops an agent-based framework to generate high-resolution contact matrices by simulating task-driven behaviour, staff workflows, and movement through shared spaces. Rather than prescribing contacts, interactions emerge from scheduled activities and proximity during task execution. The model is parameterised using activity-diary data from aged care workers and separates behavioural logic from physical layout, enabling adaptation to different facility designs without altering core mechanisms. Results show strong heterogeneity in contact patterns across care levels and staff shifts. Low and medium care residents had higher contact frequencies than high care residents, while day and afternoon staff shifts accounted for most resident-staff interactions. Contacts clustered around daily routines such as meals and communal activities. Incorporating a proximity-based airborne transmission component showed that risk was concentrated during high-contact shifts and among more mobile residents. Vaccination scenarios substantially reduced predicted transmission, with the greatest impact when both staff and residents were vaccinated. By linking organisational processes to emergent contact structure, this framework provides a reproducible approach to contact matrix generation for institutional settings, supporting more realistic transmission modelling and evaluation of targeted infection control strategies.

In the context of clinical research, computational models have received increasing attention over the past decades. In this systematic review, we aimed to provide an overview of the role of so-called in silico clinical trials (ISCTs) in medical applications. Exemplary for the broad field of clinical medicine, we focused on in silico (IS) methods applied in drug development, sometimes also referred to as model informed drug development (MIDD). We searched PubMed and ClinicalTrials.gov for published articles and registered clinical trials related to ISCTs. We identified 202 articles and 48 trials, and of these, 76 articles and 19 trials were directly linked to drug development. We extracted information from all 202 articles and 48 clinical trials and conducted a more detailed review of the methods used in the 76 articles that are connected to drug development. Regarding application, most articles and trials focused on cancer and imaging-related research while rare and pediatric diseases were only addressed in 14 articles and 5 trials, respectively. While some models were informed combining mechanistic knowledge with clinical or preclinical (in-vivo or in-vitro) data, the majority of models were fully data-driven, illustrating that clinical data is a crucial part in the process of generating synthetic data in ISCTs. Regarding reproducibility, a more detailed analysis revealed that only 24% (18 out of 76) of the articles provided an open-source implementation of the applied models, and in only 20% of the articles the generated synthetic data were publicly available. Despite the widely raised interest, we also found that it is still uncommon for ISCTs to be part of a registered clinical trial and their application is restricted to specific diseases leaving potential benefits of ISCTs not fully exploited.

Understanding the dynamic behavior of biomolecules is fundamental to elucidating biological function and facilitating drug discovery. While Molecular Dynamics (MD) simulations provide a rigorous physical basis for studying these dynamics, they remain computationally expensive for long timescales. Conversely, recent deep generative models accelerate conformation generation but are typically either failing to model temporal relationship or built only for monomeric proteins. To bridge this gap, we introduce ATMOS, a novel generative framework based on State Space Models (SSM) designed to generate atom-level MD trajectories for biomolecular systems. ATMOS integrates a Pairformer-based state transition mechanism to capture long-range temporal dependencies, with a diffusion-based module to decode trajectory frames in an autoregressive manner. ATMOS is trained across crystal structures from PDB and conformation trajectory from large-scale MD simulation datasets including mdCATH and MISATO. We demonstrate that ATMOS achieves state-of-the-art performance in generating conformation trajectories for both protein monomers and complex protein-ligand systems. By enabling efficient inference of atomic trajectory of motions, this work establishes a promising foundation for modeling biomolecular dynamics.

Travelling wave solutions of reaction-diffusion equations are widely used to model the spatial spread of populations and other phenomena in biology and physics. In this article, we reinterpret the classical variational principle approach through an optimal control formulation, in order to obtain a lower bound on the invasion speed of travelling wave solutions in systems of nonlinear partial differential equations. We begin by analysing single-species models, where the evolution of the density is governed by a scalar equation with a density-dependent diffusion term and a nonlinear reaction term. We show that for any admissible test function, maximising with respect to the parameter of interest yields a bound on the travelling wave speed. We apply this framework to several examples, including the porous-Fisher equation, and examine when nonlinear selection mechanisms dominate over the classical linear marginal stability criterion. Extending this approach, we then consider multi-species systems of reaction-diffusion equations and, reframed as Pontryagin-type optimality systems, we derive analogous bounds on the travelling wave speed using a variational framework under weak coupling. Finally, we employ numerical simulations to confirm the accuracy of the predicted wave speeds across a range of illustrative examples.

Automated Alzheimer's Disease (AD) screening has predominantly followed the inductive paradigm of pattern recognition, which directly maps the input signal to the outcome label. This paradigm sacrifices construct validity of clinical protocol for statistical shortcuts. This paper proposes Agentic Cognitive Profiling (ACP), an agentic framework that realigns automated screening with clinical protocol logic across multiple cognitive domains. Rather than learning opaque mappings from transcripts to labels, the framework decomposes standardized assessments into atomic cognitive tasks and orchestrates specialized LLM agents to extract verifiable scoring primitives. Central to our design is decoupling semantic understanding from measurement by delegating all quantification to deterministic function calling, thereby mitigating hallucination and restoring construct validity. Unlike popular datasets that typically comprise around a hundred participants under a single task, we evaluate on a clinically-annotated corpus of 402 participants across eight structured cognitive tasks spanning multiple cognitive domains. The framework achieves 90.5% score match rate in task examination and 85.3% accuracy in AD prediction, surpassing popular baselines while generating interpretable cognitive profiles grounded in behavioral evidence. This work demonstrates that construct validity and predictive performance need not be traded off, charting a path toward AD screening systems that explain rather than merely predict.

Glycans are structurally diverse and flexible biomolecules that play key roles in many biological processes. Their conformational variability makes the modeling of their interactions with proteins particularly challenging. This chapter presents a step-by-step protocol for modeling protein-glycan interactions using HADDOCK3, an integrative modeling platform that supports the inclusion of experimental or predicted interaction restraints and allows for flexible refinement of the solutions. The workflow is illustrated using the interaction between a linear homopolymer glycan, 4-beta-glucopyranose, and the catalytic domain of the Humicola grisea Cel12A enzyme, for which an experimental X-ray structure is available as a reference. Detailed instructions are provided for input structure preparation, restraint definition, docking setup, execution, and result analysis. Application of the protocol starting from unbound structures yields models of acceptable to medium quality, with interface-ligand RMSD values below 3 angstroms. Although illustrated on a specific system, the protocol has been optimized and benchmarked on multiple protein-glycan complexes and is broadly applicable to similar systems, providing a framework for integrative modeling of protein-glycan interactions.

We propose Q-BIOLAT, a framework for modeling and optimizing protein fitness landscapes in binary latent spaces. Starting from protein sequences, we leverage pretrained protein language models to obtain continuous embeddings, which are then transformed into compact binary latent representations. In this space, protein fitness is approximated using a quadratic unconstrained binary optimization (QUBO) model, enabling efficient combinatorial search via classical heuristics such as simulated annealing and genetic algorithms. On the ProteinGym benchmark, we demonstrate that Q-BIOLAT captures meaningful structure in protein fitness landscapes and enables the identification of high-fitness variants. Despite using a simple binarization scheme, our method consistently retrieves sequences whose nearest neighbors lie within the top fraction of the training fitness distribution, particularly under the strongest configurations. We further show that different optimization strategies exhibit distinct behaviors, with evolutionary search performing better in higher-dimensional latent spaces and local search remaining competitive in preserving realistic sequences. Beyond its empirical performance, Q-BIOLAT provides a natural bridge between protein representation learning and combinatorial optimization. By formulating protein fitness as a QUBO problem, our framework is directly compatible with emerging quantum annealing hardware, opening new directions for quantum-assisted protein engineering. Our implementation is publicly available at: https://github.com/HySonLab/Q-BIOLAT

Accurate diagnosis of Alzheimer's disease (AD) requires handling tabular biomarker data, yet such data are often small and incomplete, where deep learning models frequently fail to outperform classical methods. Pretrained large language models (LLMs) offer few-shot generalization, structured reasoning, and interpretable outputs, providing a powerful paradigm shift for clinical prediction. We propose TAP-GPT Tabular Alzheimer's Prediction GPT, a domain-adapted tabular LLM framework built on TableGPT2 and fine-tuned for few-shot AD classification using tabular prompts rather than plain texts. We evaluate TAP-GPT across four ADNI-derived datasets, including QT-PAD biomarkers and region-level structural MRI, amyloid PET, and tau PET for binary AD classification. Across multimodal and unimodal settings, TAP-GPT improves upon its backbone models and outperforms traditional machine learning baselines in the few-shot setting while remaining competitive with state-of-the-art general-purpose LLMs. We show that feature selection mitigates degradation in high-dimensional inputs and that TAP-GPT maintains stable performance under simulated and real-world missingness without imputation. Additionally, TAP-GPT produces structured, modality-aware reasoning aligned with established AD biology and shows greater stability under self-reflection, supporting its use in iterative multi-agent systems. To our knowledge, this is the first systematic application of a tabular-specialized LLM to multimodal biomarker-based AD prediction, demonstrating that such pretrained models can effectively address structured clinical prediction tasks and laying the foundation for tabular LLM-driven multi-agent clinical decision-support systems. The source code is publicly available on GitHub: https://github.com/sophie-kearney/TAP-GPT.

Secure communication is the cornerstone of modern infrastructures, yet achieving unconditional security -resistant to any computational attack- remains a fundamental challenge. The One-Time Pad (OTP), proven by Shannon to offer perfect secrecy, requires a shared random key as long as the message, used only once. However, distributing large keys over long distances has been impractical due to the lack of secure and scalable sharing options. Here, we introduce a DNA-based cryptographic primitive that leverages random pools of synthetic DNA to install a synchronized entropy source between distant parties. Our approach uses duplicated DNA molecules -comprising random index-payload pairs- as a shared secret. These molecules are locally sequenced and digitized to generate a common binary mask for OTP encryption, achieving unconditional security without relying on computational assumptions. We experimentally demonstrate this protocol between Tokyo and Paris, using in-house sequencing, generating a shared secret mask of $\sim$ 400 Mb with a residual error rate to achieve the usual overall decryption failure rate of $2^{-128}$. The min-entropy of the binary mask meets the most recent National Institute of Standards and Technology requirements (SP 800-90B), and is comparable to that of approved cryptographic random number generators. Critically, our system can resist two types of adversarial interference through molecular copy-number statistics, providing an additional layer of security reminiscent of Quantum Key Distribution, but without distance limitations. This work establishes DNA as a scalable entropy source for long-distance OTP, enabling high-throughput and secure communications in sensitive contexts. By bridging molecular biology and cryptography, DNA-based key distribution opens a promising new route toward unconditional security in global communication networks.

Scalable manufacturing of human induced pluripotent stem cells (iPSCs) is essential for industrial-scale production of cell therapies and regenerative medicines. However, the 3D aggregate cultures used in manufacturing exhibit substantial spatial and metabolic heterogeneity compared with the relatively homogeneous monolayer systems used in laboratory studies, complicating mechanistic understanding and predictive metabolic modeling across culture scales. To address this challenge, we developed a modular multiscale mechanistic foundation model that links molecular, cellular, and macroscopic processes while accounting for spatial and metabolic heterogeneity. The framework integrates extracellular culture dynamics, intracellular metabolic fluxes, and cellular redox states by extending a previously established monolayer kinetic network and coupling it with a biological systems-of-systems (Bio-SoS) multiscale model for aggregate cultures, incorporating explicit redox interactions. Systematic monolayer and aggregate experiments (including multiple isotopic tracers, extracellular metabolite profiling, and two-photon optical redox imaging) were used to improve and validate the model. This integrated framework unifies heterogeneous datasets across culture configurations and enables mechanistic interpretation of metabolic and redox responses across heterogeneous culture scales, providing a quantitative foundation for scalable iPSC biomanufacturing.

In this paper we expand the concept of biological speciation by symmetry breaking of Golubitsky and Stewart to the case of three clades in which N populations following the same dynamical laws can separate. The underlying differential equation is based on a fifth order polynomial of a trait variable with first or second order coupling. We present some general strategies to find all possible steady states and their stabilities. Numerical data are given for a specific system. We show the locations of three-clade distributions in dependence on the coupling and an environmental parameter. The results show a decrease of the number of stable states with higher coupling and a higher probability of ending in a three-clade state for larger N. Limits and potentials of the approach if zero roots for the trait variable occur are discussed.

Many diatoms follow a size diminuation - size restoration cycle in their vegetative phase, leading to daughter cells that differ in size. For the diatom Seminavis robusta, we investigated by cell tracking over several generations whether the size difference reflects also in different intermitotic times or in the mobility of the cells. A tracking setup and machine-learning based detection algorithm was developed that revealed no significant difference in intermitotic times, a weak coupling to the day- night cycle, and a higher motility of the hypothecal, smaller daughter cell.

Routine oncologic computed tomography (CT) presents an ideal opportunity for screening spinal instability, yet prophylactic stabilization windows are frequently missed due to the complex geometric reasoning required by the Spinal Instability Neoplastic Score (SINS). Automating SINS is fundamentally hindered by metastatic osteolysis, which induces topological ambiguity that confounds standard segmentation and black-box AI. We propose Topology-Guided Biomechanical Profiling (TGBP), an auditable white-box framework decoupling anatomical perception from structural reasoning. TGBP anchors SINS assessment on two deterministic geometric innovations: (i) canal-referenced partitioning to resolve posterolateral boundary ambiguity, and (ii) context-aware morphometric normalization via covariance-based oriented bounding boxes (OBB) to quantify vertebral collapse. Integrated with auxiliary radiomic and large language model (LLM) modules, TGBP provides an end-to-end, interpretable SINS evaluation. Validated on a multi-center, multi-cancer cohort ($N=482$), TGBP achieved 90.2\% accuracy in 3-tier stability triage. In a blinded reader study ($N=30$), TGBP significantly outperformed medical oncologists on complex structural features ($κ=0.857$ vs.\ $0.570$) and prevented compounding errors in Total Score estimation ($κ=0.625$ vs.\ $0.207$), democratizing expert-level opportunistic screening.

Raman spectroscopy is a promising tool for microbial identification, yet its implementation in microbiology and clinical workflow is still restricted due to the accompanying additional preparation required to focus on microbial signals. Here, we demonstrate Raman-based bacterial identification directly from unopened, inverted agar plates, the same conditions used during incubation. Our approach enabled identification with single gene-level sensitivity using two Escherichia coli variants, differing only in green fluorescent protein (GFP) expression, across diverse media and substrate material conditions, despite the interrogation path traversing 3-4 mm thick background material. We integrated traditional density functional theory (DFT)-based material computation with machine learning analysis, achieving over 97.7% classification accuracy, surpassing the performance of standard measurements from opened plates by 10.8% higher mean accuracy and 0.76% less variance. We further demonstrated Raman mapping-based colony identification via Raman peaks characteristic to GFPmut3 chromophore structure generated by DFT. Our approach is robust to changes in algorithms or substrate materials and promises real-time, non-perturbative monitoring of bacterial growth, biofilm formation, and antimicrobial resistance development.

Ultrasound imaging is an essential first-line tool for assessing hepatic steatosis. While conventional B-mode ultrasound imaging has limitations in providing detailed tissue characterization, ultrasound Nakagami imaging holds promise for visualizing and quantifying tissue scattering in backscattered signals, with potential applications in fat fraction analysis. However, existing methods for Nakagami imaging struggle with optimal window size selection and suffer from estimator instability, leading to degraded image resolution. To address these challenges, we propose a novel method called UNICORN (Ultrasound Nakagami Imaging via Score Matching and Adaptation), which offers an accurate, closed-form estimator for Nakagami parameter estimation based on the score function of the ultrasound envelope signal. Unlike methods that visualize only specific regions of interest (ROI) and estimate parameters within fixed window sizes, our approach provides comprehensive parameter mapping by providing a pixel-by-pixel estimator, resulting in high-resolution imaging. We demonstrated that our proposed estimator effectively assesses hepatic steatosis and provides visual distinction in the backscattered statistics associated with this condition. Through extensive experiments using real envelope data from patient, we validated that UNICORN enables clinical detection of hepatic steatosis and exhibits robustness and generalizability.

Atolls are traditionally explained as the result of coral reefs accreting around volcanic islands followed by gradual subsidence, yielding a hollow, ring-shaped rim that can extend for kilometres. However, satellite imagery shows that similar annular outlines also appear in much smaller patch reefs, where atoll-forming geological pathways do not apply. In some systems, small annular patches occur within the lagoons of larger atolls, producing nested ring-like patterns. The recurrence of annularity across such contrasting contexts and scales suggests that shared, self-organising processes may also contribute to shaping these reefs. Here, we test whether interactions between reef growth and marine currents can generate annular forms and explain their cross-scale geometric regularities. We develop a numerical model in which coral growth follows simple process-based rules, with local colonisation and mortality depending on resource supply and hydrodynamic stress, and water flow resolved using fluid dynamics. Simulations show that this coupling robustly produces ring-like patch reefs and atoll-like configurations across spatial scales, consistent with observed morphologies. Beyond qualitative agreement, the emergent reefs reproduce key geometric signatures reported in global datasets, including scaling laws and fractal dimensions. Together, these results identify coral-current interactions as a plausible pathway to annular reef formation and a mechanistic explanation for scale-free reef geometry.

Large language models (LLMs) are becoming an increasingly important component of human--computer interaction, enabling users to coordinate a wide range of intelligent agents through natural language. While language-based interfaces are powerful and flexible, they implicitly assume that users can reliably produce explicit linguistic input, an assumption that may not hold for users with speech or motor impairments, e.g., Amyotrophic Lateral Sclerosis (ALS). In this work, we investigate whether neural signals can be used as an alternative input to LLMs, particularly to support those socially marginalized or underserved users. We build a simple brain-LLM interface, which uses EEG signals to guide image generation models at test time. Specifically, we first train a classifier to estimate user satisfaction from EEG signals. Its predictions are then incorporated into a test-time scaling (TTS) framework that dynamically adapts model inference using neural feedback collected during user evaluation. The experiments show that EEG can predict user satisfaction, suggesting that neural activity carries information on real-time preference inference. These findings provide a first step toward integrating neural feedback into adaptive language-model inference, and hopefully open up new possibilities for future research on adaptive LLM interaction.

This study presents a mathematical model that captures the interactions among tumor cells, healthy cells, and immune cells in a tumor-bearing host, with a specific focus on breast cancer. Incorporating the concept of delay, the model consists of four differential equations to analyze these cellular dynamics. The findings demonstrate the superior efficacy of metronomic chemotherapy compared to the maximum tolerated dose (MTD) method and underscore the necessity of adjunct therapies. Oscillatory tumor cell dynamics revealed by the model highlight the challenges of achieving complete tumor elimination through chemotherapy alone. Sensitivity analysis confirms the robustness of the model, particularly under metronomic treatment protocols, aligning with experimental observations regarding metronomic-to-MTD dosage ratios. Furthermore, the results emphasize the importance of synergistic effects from combination therapies. This biologically consistent framework provides valuable insights into tumor-immune interactions and offers a foundation for optimizing therapeutic strategies in cancer treatment.

This work introduces a framework for reconstructing the interaction graph of neuronal networks modeled as multivariate point processes. The methodology performs bivariate inference, identifying synaptic links exclusively from the spike trains of a pair of neurons, without requiring observations of the remaining network activity. We propose a Macro-Micro Extrapolation algorithm to address data sparsity at the micro-scale, inferring synaptic interactions in the limit $Δ\to 0^+$. A key contribution is the Spike-Triggered Estimator, which leverages the local reset property of Galves-Löcherbach dynamics to decouple local synaptic jumps from higher-order network contributions, significantly reducing estimation variance and eliminating spurious dependencies on baseline firing intensities. By employing an adaptive hybrid logic that switches between sample averaging and our novel Pyramid Extrapolation, we ensure robust classification of excitatory, inhibitory, and null connections even in low signal-to-noise regimes. The framework's scalability and precision are validated by numerical results on dense cliques and structured layered networks, achieving perfect classification accuracy across diverse topological motifs.

Biomedical knowledge is fragmented across siloed databases -- Reactome for pathways, STRING for protein interactions, ClinicalTrials.gov for study registries, DrugBank for drug vocabularies, DGIdb for drug-gene interactions, SIDER for side effects. We present three open-source biomedical knowledge graphs -- Pathways KG (118,686 nodes, 834,785 edges from 5 sources), Clinical Trials KG (7,774,446 nodes, 26,973,997 edges from 5 sources), and Drug Interactions KG (32,726 nodes, 191,970 edges from 3 sources) -- built on Samyama, a high-performance graph database written in Rust. Our contributions are threefold. First, we describe a reproducible ETL pattern for constructing large-scale KGs from heterogeneous public data sources, with cross-source deduplication, batch loading (Python Cypher and Rust native loaders), and portable snapshot export. Second, we demonstrate cross-KG federation: loading all three snapshots into a single graph tenant enables property-based joins across datasets. Third, we introduce schema-driven MCP server generation for LLM agent access, evaluated on a new BiomedQA benchmark (40 pharmacology questions): domain-specific MCP tools achieve 98% accuracy vs. 85% for schema-aware text-to-Cypher and 75% for standalone GPT-4o, with zero schema errors. All data sources are open-license. The combined federated graph (7.9M nodes, 28M edges) loads in approximately 3 minutes on commodity cloud hardware, with single-KG queries completing in 80-100ms and cross-KG federation joins in 1-4s

The unprecedented extension of the human lifespan necessitates a parallel evolution in how we quantify the quality of aging and its socioeconomic impact. Traditional metrics focusing on Healthspan (years free of disease) overlook the gradual erosion of physiological capacity that occurs even in the absence of illness, leading to declines in productivity and eventual lack of capacity to work. To address this critical gap, we introduce Peakspan: the age interval during which an individual maintains at least 90% of their peak functional performance in a specific physiological or cognitive domain. Our multi-system analysis reveals a profound misalignment: most biological systems reach maximal capacity in early adulthood, resulting in a Peakspan that is remarkably short relative to the total lifespan. This dissociation means humans now spend the majority of their adult lives in a "healthy but declined" state, characterized by a significant functional gap. We argue that extending Peakspan and developing strategies to restore function in post-peak individuals is the functional manifestation of rejuvenative biomedical progress and is essential for sustained economic growth in aging societies. Recognizing and tracking Peakspan, increasingly facilitated by artificial intelligence and foundational models of biological aging, is crucial for developing strategies to compress functional morbidity and maximize human potential across the life course.

In-context learning (ICL) -- the capacity of a model to infer and apply abstract patterns from examples provided within its input -- has been extensively studied in large language models trained for next-token prediction on human text. In fact, prior work often attributes this emergent behavior to distinctive statistical properties in human language. This raises a fundamental question: can ICL arise organically in other sequence domains purely through large-scale predictive training? To explore this, we turn to genomic sequences, an alternative symbolic domain rich in statistical structure. Specifically, we study the Evo2 genomic model, trained predominantly on next-nucleotide (A/T/C/G) prediction, at a scale comparable to mid-sized LLMs. We develop a controlled experimental framework comprising symbolic reasoning tasks instantiated in both linguistic and genomic forms, enabling direct comparison of ICL across genomic and linguistic models. Our results show that genomic models, like their linguistic counterparts, exhibit log-linear gains in pattern induction as the number of in-context demonstrations increases. To the best of our knowledge, this is the first evidence of organically emergent ICL in genomic sequences, supporting the hypothesis that ICL arises as a consequence of large-scale predictive modeling over rich data. These findings extend emergent meta-learning beyond language, pointing toward a unified, modality-agnostic view of in-context learning.

Recent advances in artificial neural networks for machine learning, and language modeling in particular, have established a family of recurrent neural network (RNN) architectures that, unlike conventional RNNs with vector-form hidden states, use two-dimensional (2D) matrix-form hidden states. Such 2D-state RNNs, known as Fast Weight Programmers (FWPs), can be interpreted as a neural network whose synaptic weights (called fast weights) dynamically change over time as a function of input observations, and serve as short-term memory storage; corresponding synaptic weight modifications are controlled or programmed by another network (the programmer) whose parameters are trained (e.g., by gradient descent). In this Primer, we review the technical foundations of FWPs, their computational characteristics, and their connections to transformers and state space models. We also discuss connections between FWPs and models of synaptic plasticity in the brain, suggesting a convergence of natural and artificial intelligence.

Protein fitness landscapes frequently exhibit epistasis, where the effect of a mutation depends on the genetic context in which it occurs, i.e., the rest of the protein sequence. Epistasis increases landscape complexity, often resulting in multiple fitness peaks. In its simplest form, known as global epistasis, fitness is modeled as a non-linear function of an underlying additive trait. In contrast, more complex epistasis arises from a network of (pairwise or many-body) interactions between residues, which cannot be removed by a single non-linear transformation. Recent studies have explored how global and network epistasis contribute to the emergence of functional bottlenecks - fitness landscape topologies where two broad high-fitness basins, representing distinct phenotypes, are separated by a bottleneck that can only be crossed via one or a few mutational paths. Here, we introduce and analyze a stylized model of global epistasis with an additive underlying trait. We demonstrate that functional bottlenecks arise with high probability if the model is properly calibrated. Furthermore, our results underscore that a proper balance between neutral and non-neutral mutations is needed for the emergence of functional bottlenecks.

Comparing information structures in between deep neural networks (DNNs) and the human brain has become a key method for exploring their similarities and differences. Recent research has shown better alignment of vision-language DNN models, such as CLIP, with the activity of the human ventral occipitotemporal cortex (VOTC) than earlier vision models, supporting the idea that language modulates human visual perception. However, interpreting the results from such comparisons is inherently limited due to the "black box" nature of DNNs. To address this, we combined model-brain fitness analyses with human brain lesion data to examine how disrupting the communication pathway between the visual and language systems causally affects the ability of vision-language DNNs to explain the activity of the VOTC. Across four diverse datasets, CLIP consistently captured unique variance in VOTC neural representations, relative to both label-supervised (ResNet) and unsupervised (MoCo) models. This advantage tended to be left-lateralized at the group level, aligning with the human language network. Analyses of 33 stroke patients revealed that reduced white matter integrity between the VOTC and the language region in the left angular gyrus was correlated with decreased CLIP-brain correspondence and increased MoCo-brain correspondence, indicating a dynamic influence of language processing on the activity of the VOTC. These findings support the integration of language modulation in neurocognitive models of human vision, reinforcing concepts from vision-language DNN models. The sensitivity of model-brain similarity to specific brain lesions demonstrates that leveraging manipulation of the human brain is a promising framework for evaluating and developing brain-like computer models.

We investigate a large ensemble of Quadratic Integrate-and-Fire (QIF) neurons with heterogeneous input currents and adaptation variables. Our analysis reveals that for a specific class of adaptation, termed quadratic spike-frequency adaptation (QSFA), the high-dimensional system can be exactly reduced to a low-dimensional system of ordinary differential equations, which describes the dynamics of three mean-field variables: the population's firing rate, the mean membrane potential, and a mean adaptation variable. The resulting low-dimensional firing rate equations (FRE) uncover a key generic feature of heterogeneous networks with spike frequency adaptation: Both the center and the width of the distribution of the neurons' firing frequencies are reduced, and this largely promotes the emergence of collective synchronization in the network. Our findings are further supported by the bifurcation analysis of the FRE, which accurately captures the collective dynamics of the spiking neuron network, including phenomena such as collective oscillations, bursting, and macroscopic chaos.

Animals flexibly change their behavior depending on context. It is reported that the hippocampus is one of the most prominent regions for contextual behaviors, and its sequential activity shows context dependency. However, how such context-dependent sequential activity is established through reorganization of neuronal activity (remapping) is unclear. To better understand the formation of hippocampal activity and its contribution to context-dependent flexible behavior, we present a novel biologically plausible reinforcement learning model. In this model, Context selector promotes the formation of context-dependent sequential activity and allows for flexible switching of behavior in multiple contexts. This model reproduces a variety of findings from neural activity, optogenetic inactivation, human fMRI, and clinical research. Furthermore, our model predicts that imbalances in the ratio between sensory and contextual representations in Context selector account for schizophrenia (SZ) and autism spectrum disorder (ASD)-like behaviors.