900 worms of C
Resampled calcium traces and consensus connectomes let labs compare activity across animals and studies.
· “Homogenized textit{C. elegans} Neural Activity and Connectivity Data”
q-bio.NC
Neurons and Cognition
Synapse, cortex, neuronal dynamics, neural network, sensorimotor control, behavior, attention
sort pith recommended most recent
Lorentzian family is unique invariant under Riccati transport
Reformulating dynamics on the circle shows the Cauchy law is the sole rotation-invariant measure, unifying exact mean-field reductions.
· “Geometric Origin of Exact Mean-Field Reductions: M{\"o}bius Symmetry and the Lorentzian Ansatz”
Zebrafish schooling is driven by spinal bend sensors
A curvature-sensing spinal loop times tail beats to a neighbor's wake; removing it dissolves the school.
Frequent AI use does not mean AI literacy
On a new 18-item GenAI exam, teens who used chatbots most often scored lowest, while perceived usefulness tracked nothing.
Video AI misses how primate brains separate motion from appearance
Temporal integration alone fails; only predictive world models approach the brain's appearance-invariant motion coding.
· “Primate vision reveals a missing principle for robust dynamic AI”
Dendritic branches are necessary for powerful local learning
Without compartments, synapses lack local error signals for deep learning without separate phases.
Researchers measured brain waves while people tapped along with a visual metronome
Filling blank intervals in a visual metronome with a colored disk improved tapping stability and was associated with reduced visual…
· “Temporal filling-in reduces attentional fluctuations in sustained visual attention”
One noisy exposure trains both replay directions
A delay-coupled ring model replays forward and backward at quantized speeds, without symmetric plasticity.
· “Forward and reverse delay-driven hippocampal replay without symmetric plasticity”
Tool-selection accuracy jumps from 23% to 94% with a research harness
Governed agentic neuroimaging: sealed plans, audited claims, and verified evidence grounding up nearly fivefold.
Grid cells cut spatial aliasing by up to 99 percent
A periodic internal spatial code disambiguates visually identical locations that boundary cues alone confuse.
· “The Role of Grid Cells in Reducing Spatial Aliasing in Hippocampal Place Representations”
Q-networks turn first-person experience into category theory
Counting connections and loops in a network of possible experiences yields structural markers for unity, self, and time.
71.2% peak accuracy: Riemannian hypergraph fusion tops cross-day BCI
Method adapts to streaming EEG and ECoG without labels, beating all tested baselines on three four-class datasets.
Global workspace theory gets a testable criterion
A four-part dynamical signature locates mediating subnetworks and separates gain from organization in real brain recordings.
· “A Control-Theoretic Formulation of Global Workspace Theory”
Reciprocal cortical loops preserve anticipated synchronization
A two-population model shows receiver-leads-sender phases and bistable switching survive bidirectional coupling.
90-degree fall doubles oxidative stress in a neuron-packed head
A 90-degree drop created the clearest injury signal, doubling oxidative stress and causing focal cell drift.
Five-layer knowledge states make LLM research knowledge shareable
Valhalla wraps files, entities, and relations in stable layers so teams can hand off knowledge states, not just prompts.
Pulse phase alone decides whether a neural network syncs or desyncs
Repeated pulses aimed at the amplitude-suppressing phase gradually weaken synchronized oscillations.
Synaptic delay slows gamma waves and sharpens population synchrony
In a simulated PING network, longer delays also reshape how transient inputs alter oscillation timing and amplitude.
Four toolkits converge toward personalized brain health
Statistics, geometry, AI, and dynamics all aim at one goal: early, individual-specific detection.
· “Data-driven techniques for translational neuroscience and personalized neuro-health”
Brain activity redirects where tau spreads in Alzheimer's
Epidemic model: firing neurons shift the threshold and course of pathology, and PET scans back the prediction.
Post-endpoint randomisation makes 'the past explains it' testable
A delay drawn after the endpoint is fixed becomes a probe; a surviving ordering rejects past-adapted sufficiency.
Structural alignment grades how much of another mind we can know
The paper maps two structural theories of consciousness onto a four-level categorical ladder from complete dictionary to no inference.
· “The Rosetta Stone and Levels of Principled Inference to the Experience of Another Mind”
EEG synchrony identifies visual learners across subjects at 70%
Phase Locking Value beats power and entropy features for cross-subject learning-style recognition, but only when brain networks differ.
Language model conflict effect comes from two competing pathways
When a prompt rule clashes with the stored same-color default, the two routes compete and incongruent trials suffer.
Space, time, and objects are causal structure itself
Each felt spot, moment, and object maps one-to-one to brain causal structure, yielding testable predictions.
· “Consciousness as Intrinsic Structure: Towards a Chemistry of Experience”
Molecular tweaks can flip whole-brain activity into anesthesia slow waves
A biophysical mean-field chain links synaptic receptors to whole-brain dynamics, validated on anesthesia.
· “A class of mean-field models to bridge molecular to brain scales”
3-5 minutes of data fix misleading neural-tracking correlations
A Fisher-transform null model estimates significance at any window length and ranks EEG features on one scale.
Quality factor caps how many labels a biological oscillator can carry
The same bound, M ≤ Q, rules out the proposed 30 GHz cortical microwave label and leaves low-frequency rhythms standing.
Generalized cross-validation sharpens brain-distance estimates
Adding within-partition between-condition distances yields more accurate Euclidean and correlation estimates in MEG pattern analysis.
· “Improved cross-validated distances for multivariate pattern analysis”
Brain fatigue may be a drop in ATP free energy
Model links falling ATP energy to shorter-lived protein solitons and hyperexcitable neurons.
· “Reduced Gibbs free energy supply hinders brain information processing during mental fatigue”
Node centrality in inferred maps predicts firing
Graph-theory scores computed on reservoir-computed connectivity track spiking, bursting, and interval statistics electrode by electrode.
· “Graph Analysis of Neuronal-Culture Connectivity Derived from a Reservoir-Computing Model”
Prior walking biases later choices even though accuracy holds
The same person could land on different sides of a clinical threshold depending on what task came first.
One energy landscape unifies vision
The same predictive-coding machinery yields surprisal, N400/P600 effects, bistable perception, and falsifiable contrasts with transformers.
· “A Hierarchical Energy-Based Model for Multimodal Cognition”
FedDOSE beats seven federated baselines on ASD and ADHD scans
Separating disease, phenotype, and scanner signals gets multi-site accuracy within 2 percent of centralized training.
Model trained on UK data finds Alzheimer's atrophy in Chinese scans
Fine-tuning lifts the detection score to 0.848; using the model as-is still hits 0.815.
· “International Transfer of Stochastic Cortical Self-Reconstruction”
Rare-category bias persists without feedback or labels
Two experiments strip the task bare and the bias survives; one auto-associative model unifies all three procedures.
Class clusters emerge and sharpen in label-free deep networks
Deeper unsupervised layers sharpen the class boundaries hidden in the data, until a few hard pairs erode accuracy.
· “Convergent Evolution in Neural Representation Space: Emergent Order in Deep Belief Networks”
Same-task neural nets converge onto shared weight regions
Matching hidden neurons by permutation shows the task, not chance, decides where training lands in weight space.
Aging stretches brain activity
EEG geometry links richer neural codes to less stability, a possible early signpost for MCI and Alzheimer's.
· “Complexity and Stability of Neural Activity Across Aging and Neurodegenerative Disease”
One equation splits base-rate neglect into two independent weights
The cue's own frequency is a second correction learners can skip, and a rating experiment can separate it.
· “Two base rates, two weights: base-rate neglect has a second axis”
Spinal stimulation disrupts ankle sense
Walking narrows measurably under stimulation, and feedback training reverses the sensory disruption.
Surprise cascades can emerge from local contrastive learning alone
Inverted Forward-Forward model cancels expected inputs bottom-up, mirroring mouse visual cortex dynamics.
· “From Local Learning to Global Prediction Through Layered Surprise Cascades”
Noise-based pruning beats weight-size pruning in trained RNNs
A local, unsupervised rule keeps more task accuracy through 80% pruning, rivaling costlier second-order methods.
New taxonomy sorts insomnia by which sleep mechanism fails
Which sleep operation fails, at which boundary, and why—a three-axis framework for subtyping insomnia.
Machines lack ethical judgment because they have nothing at stake
AI ethics is human ethics: ground it in the brain's reward cycle and global workspace, not in new machine rules
Sameness in autism may be an entropy-minimizing strategy
A new model treats rigid routines as the only way to cut surprise when abstract learning is closed off.
· “An entropic explanation of insistence on sameness in autism”
Visual perception has two time axes
Separating stimulus and processing time lets one experiment test oscillation, prediction, and coarse-to-fine accounts.
· “Time²: A framework for the neural dynamics of visual perception”
Topology-picked driver nodes broaden brain control at equal cost
Choosing drivers by persistent cycles, not degree, changes which brain states are cheap to reach—at no extra energy.
· “Persistent homology broadens the controllable subspace in human structural connectomes”
Sleep-stage shifts may be fold
One noisy neural-field potential generates all four signature classes, each with its own EEG test.
· “A Landau-Ginzburg Phenomenology of Sleep-Stage Transitions”
Remission is the most likely next step after childhood suicidal thoughts
NSSI makes trajectories riskier and less predictable, so apparent recovery deserves continued monitoring.
High-frequency EEG modes hit 98% in alcohol-vs-control split
A Rayleigh filter keeps only non-random trials; the trial-level split still leaves cross-subject generalization untested.
· “Detecting high-frequency brain disorder signals using dynamic mode decomposition from EEG”
Demand is a mathematical inevitability
A set-theoretic derivation shows that any deterministic planner already maximizes its own output as the standard
Brain world model forecasts fMRI without access to future stimuli
Two-stage latent model predicts whole-brain fMRI up to 100 TRs ahead and beats regression baselines on 3 benchmarks.
· “NeuroWorld: A Latent Brain World Model for Stimulus-Conditioned Human Brain Dynamics”
39.8% speech retrieval, 20x smaller, and readable
The decoder's weights map to cortical sources; occlusion shows silence, loudness, vowels, and onsets drive it.
Capacity loss is the metric tensor
Treating weight distributions as geodesic flow improves generalization and mitigates barren plateaus.
· “Statistical Mechanics of Learning on Product Wasserstein Manifolds”
Hippocampal replay works like data augmentation
A perspective turns theories of memory into models that learn from the same sensory input as animals.
· “Data augmentation as a framework for modeling hippocampal contributions to generalization”
Hybrid neural net hits 86.9% accuracy reading Fragile X EEGs
Combining alpha and gamma bands with recurrence plots beats single-signal models on held-out subjects.
· “Deep Learning CNN and Recurrence Analysis for Alpha Gamma EEG Biomarkers in Fragile X Syndrome”
One shared Gaussian process implements the Bayesian brain
A formal bridge maps a single GP with layer-index input onto the cortical microcircuit's predictions and prediction errors.
Receptor effects survive mean-field reduction to whole-brain dynamics
Master-equation mean fields keep synaptic kinetics and adaptation manipulable across scales, provided every reduction step is validated.
Eight of nine VLMs match different adult groups across ToM tasks
One model can look adult on perspective-taking but not intention attribution; single benchmarks miss the split.
· “Cross-Task Dissociation in Frontier Vision-Language Model Theory of Mind”
No linear time-invariant system can habituate
Nonlinearity is a structural requirement; one leaky memory plus a static nonlinear readout suffices for the core hallmarks.
· “Dynamical principles of habituation across substrates and scales”
A fly's compass update is its network's cheapest move
Control-theoretic cost landscape also explains sensory vs association cortex and how training sculpts neural networks.
Depression model adapts to 7 new brain-scan sites at 73.6% accuracy
Jointly aligning three fMRI connectivity views and two source sites beats single-source baselines by nearly six points.
Depletion accelerates rate learning but not composition learning
Falling encounter intervals reveal a patch's initial yield, while the mix of patch types is learned one visit at a time.
· “Resource depletion accelerates rate learning but not composition learning in patch foraging”
One Python function computes all 17 metacognition measures
Non-model-based measures match the reference pipeline to numerical precision, so labs can switch without losing comparability.
· “metasignal: A Python Package for Comprehensive Metacognitive Analysis and Decision-Making”
Flicker fusion marks the line between plastic and fixed vision
A stable perceptual clock would explain why temporal acuity resists training and may limit working memory.
For mice and humans, 'natural behavior' does not exist
Neuroscience should pick behaviors by survival and reproduction, not naturalness
ToM plus cost comparison decides when to watch others
A utility model that mentalizes about goals matches human observe-or-explore choices across four games
· “Using Theory of Mind to Arbitrate between Social and Non-social Learning”
Daily shocks shrink organoid responses from 93% to 10%
A first-stimulation control at matched age shows the collapse is from stimulation history, not maturation
Adaptive brain-community GNN tops Alzheimer's fMRI classification
Learned communities match canonical brain networks and expose a dedifferentiation pattern in Alzheimer's disease.
· “MPP-GNN: Subject-Adaptive Community Detection for fMRI-Based Alzheimer's Disease Classification”
One EEG model repairs any channel layout
A 380M-parameter diffusion autoencoder matches a fixed-window specialist and beats spherical spline interpolation on corrupted EEG.
· “ZUNA1.1: A more flexible EEG foundation model for Denoising and Super-resolution”
86% of AI brain-stimulation studies stay at early readiness
Review of 239 studies finds external validation in only 5.9% and no system in routine clinical use.
Design is communication: mentalizing model beats efficiency baselines
A cooperative-game model predicts human design placements at r≈0.95, versus 0.32 for efficiency-only baselines.
Pain location's low diagnostic value is three problems
Three failure modes need different fixes; a fourth hidden error sits in the report itself.
A two-way information loop predicts navigation success across species
One metric built from reactive and active information flow ranks performance in bacteria, worms, flies, and AI agents.
· “A behavior-environment information loop drives sensory navigation”
LLMs and brains run on the same five cognitive principles
Human and machine intelligence converge on shared inference, architecture, representation, learning, and control.
· “Cognitive Convergence: Deep Similarities Between Large Language Models and Human Cognition”
Why polishing AI outputs adds no scientific evidence
When convincing fakes become cheap, the share of true published findings falls back to the base rate.
· “Phantom Evidence: How and Why Generative AI Manufactures False Positives in Science”
GraphIDyOM matches Lisp IDyOM to under 0.003 bits
A Python reimplementation makes the model's learned memories inspectable as graphs for analysis and interaction.
· “GraphIDyOM: A graph-native Python reimplementation of IDyOM for musical expectation modelling”
Simulated chart readers fail in human-like ways
Fast and Slow agents predict tick bias vs memory decay before you run the user study
· “Beyond the Post Hoc User Study: Modeling Visual Decision-Making with Active Inference”
EEG agent routes language, but code alone can release claims
Commit-then-confirm harness blocks fishing and lifecycle reuse while still answering registered ERP questions
