Papers

Researchers at Northwestern Polytechnical University propose SEDAT, a hybrid tokenizer for large EEG foundation models that folds squeeze-and-excitation spatial aggregation, data-adaptive Gaussian average filtering, instantaneous-frequency-guided segmentation and Fourier-domain resampling into a single efficient pipeline. Evaluated on 10 heterogeneous EEG datasets with four foundation models, LaBraM, EEGFormer, EEGPT and NeuroGPT, SEDAT improved classification by up to 15.3% over fixed-length windowing and by 1.2-4.6% over the next-best tokenizer, the authors report. The study was published in the Journal of Neural Engineering on September 1, 2026.

A research team has posted a preprint on arXiv proposing EEG-AS, an instance-level algorithm selection framework that moves the choice of an EEG foundation model down to the level of a single recording. Recent EEG foundation models perform strongly across neural decoding tasks, but none is consistently best across datasets or individual instances, and instance-level model selection has been largely unexplored. EEG-AS characterizes each instance using inference-available latent EEG embeddings, handcrafted neurophysiological features and an anchor foundation model. In training it learns to reconstruct the behaviors of models it cannot observe from privileged prediction tokens; at inference it estimates those behaviors directly, so it can choose among 7 EEG foundation models without executing the whole portfolio. In experiments on seven public EEG benchmarks the framework narrows the gap between the single best solver and the per-instance oracle upper bound, the authors say. The work is a preprint and has not been peer reviewed.

A researcher warns that AI-assisted brain-computer interfaces may over-optimize short-term proxies of success and drift from users' durable goals, a closed-loop failure mode she names neuroadaptive overfitting. Artificial intelligence is turning BCIs from task-specific neural decoders into adaptive systems that complete language, smooth movement, regulate rehabilitation support and adjust stimulation. Her Slow-Fast framework paces AI assistance according to decoder evidence, uncertainty, clinical stakes, fatigue and user-defined goals, distinguishing fast, guarded and slow assistance across communication, motor control, neurorehabilitation and closed-loop neuromodulation.

A team in France and the United States has released NETBCI, a longitudinal multimodal dataset pairing magnetoencephalography (MEG) and electroencephalography (EEG) from 19 healthy subjects across 4 sessions performed on 4 different days, built to study how brain networks reorganize during brain-computer interface training. Controlling a BCI remains a learned skill that a non-negligible proportion of users never acquire even after several sessions, the authors write, and the causes of that inter-individual variability remain an open question. Each session comprises 2 eyes-open resting-state recordings of 3 minutes each plus 6 runs in which participants either sustained right-hand motor imagery or stayed at rest to control the position of a virtual cursor, and the release also includes anonymized MRI scans and behavioral scores; the authors say they hope the sample size and range of modalities will support analyses beyond brain network reorganization.

A chip-level design for a one-way impulse-radio ultra-wideband (IR-UWB) uplink brings the implant's packet engine and free-running transmitter, a tissue-proxy channel, and the wearable's mixed-signal receiver and digital baseband under a single layered protocol. It aims to keep circuit partitioning, waveform mapping, packet handling and failure behavior consistent for high-rate neural telemetry. The work is a preprint and has not been peer reviewed.

Researchers recording EEG from 17 basketball players and 16 novices found that only the players showed a significant, sustained weakening of sensorimotor rhythms when imagining a shot or lifting a box, most clearly for an imagined free throw. This rhythm weakens during imagined movement and is commonly used as a marker of motor cortex engagement. Among the players, however, stronger imagery ability went with a smaller weakening, which may mean the brain was working more efficiently. The researchers conclude that neurofeedback training based on brain signals should not use one-size-fits-all targets but should account for a trainee's motor expertise and imagery ability. The study is a preprint and has not been peer reviewed.

Researchers have released an open, near whole-head functional near-infrared spectroscopy (fNIRS) dataset from 57 participants and found that signal quality partly explains why brain activation differs from person to person. fNIRS, a non-invasive technique that shines light through the scalp to measure changes in blood oxygen in the brain, often yields results that vary widely between individuals and are hard to reproduce, and researchers suspect systemic physiological fluctuations and differences in signal quality may distort task-evoked hemodynamic responses. The team recorded participants at rest and during real movement, imagined movement, emotion recognition, visual and auditory tasks, alongside peripheral physiology including heart rate, respiration, skin conductance, ECG and EMG, plus short-channel recordings. Four quality metrics, including the scalp coupling index and signal-to-noise ratio, were all significantly related to channel length and to task activation estimates; group-level analysis confirmed activation in the expected brain regions. The study is a preprint and has not been peer reviewed.

Researchers put five machine-learning classifiers through a single preprocessing and common spatial pattern pipeline on three public EEG datasets and found the same algorithm's accuracy swinging from 37.50% to 60.23% depending on the dataset. Linear discriminant analysis reached 60.23% on the PhysioNet EEG Motor Movement/Imagery set, while random forest managed 55.36% on BCI Competition IV Dataset 2a under five-fold cross-validation. The authors attribute the spread to dataset characteristics, subject differences and evaluation parameters, and say it exposes a persistent comparability gap in BCI decoding research; the study appeared in the Journal of Computers, Mechanical and Management on August 31, 2026.

Researchers report that more than 150 children worldwide have received implanted brain-computer interfaces, a number expected to grow quickly as the devices reach the market. A 2026 paper in Neurorehabilitation and Neural Repair spells out a roadmap for pediatric iBCI development, including work from the first International Virtual Summit on Implanted BCIs for Children with Complex Needs and workshops at the 11th International BCI Society Meeting. It highlights unresolved questions about early-life implantation and pediatric indications that must be answered before the technology is deployed widely in children.

Implantable brain-computer interfaces are shifting from rigid silicon architectures to soft, structurally adaptive systems built for seamless, long-term integration with neural tissue, according to a review of flexible electrode materials and structural design published in SmartMat on August 31, 2026. Breakthroughs in materials science and micro/nanofabrication have given this generation of devices mechanical compliance, robust interfacial adhesion and high-fidelity signal acquisition that earlier designs could not reach, the review says. Long-term stability at the electrode-tissue interface remains one of the core bottlenecks for invasive BCI.

A team from the University of Utah, Kangwon National University and the University of Georgia describes a silicon-free route to neural microelectrode arrays: soft polymeric 3D microneedles are printed by two-photon polymerization, then turned into electrodes by injecting liquid gallium into the hollow channels. A newly defined retention number predicts whether the gallium stays put under physiological conditions. Coating the surface with gold nanoparticles and PEDOT doped with tetrafluoroborate cut impedance by roughly 3 orders of magnitude, and in vivo recordings in an invertebrate model captured neural spikes with no gallium leakage. The study was published in ACS Sensors on August 31, 2026.

Researchers at the University of Colorado Boulder have built ERP-XTTN, a cross-attention model that classifies event-related potentials (ERPs) in users it has never seen, with no per-user calibration. Across three public datasets and eight ERP components, it trailed the best baseline by 0.025 AUROC on average using only three channels. The work appeared in the Journal of Neural Engineering.

Researchers at the Second Affiliated Hospital of Xi'an Jiaotong University in northwestern China and the Key Laboratory of Biomedical Information Engineering of the Ministry of Education have built an in vitro screening platform combining electrospun PLCL with inkjet-printed reduced graphene oxide (rGO) and growth-factor micropatterns. Because it varies conductive, biochemical and topographical cues together, the platform can evaluate printing parameters and electric-field strength in a single system. Under 150 mV/cm direct-current stimulation, PC-12 cells showed more neurite-like outgrowth and better alignment than with no stimulation or at 300 mV/cm.

A Mayo Clinic team integrated CorTec's Brain Interchange (BIC), an implantable neural sensing and stimulation device, with the open-source BCI2000 software platform to create an open research system for developing and testing implantable brain-computer interfaces and closed-loop neuromodulation, in which the device reads brain signals and adjusts stimulation accordingly. In long-term validation in 5 dogs, with the longest implant exceeding 3 years, the system maintained stable chronic recording, documented gradual degradation of some channels and its mechanical causes, and still decoded brain function more than 2 years after implantation. The team also showed one-dimensional cursor control was feasible in 1 human participant and released its surgical, imaging and analysis protocols and data. The work is a preprint and has not been peer reviewed.

A team at Kunming University of Science and Technology, in southwestern China's Yunnan province, has published a paper in the Journal of Biomedical Engineering analyzing how inherent limitations set the capability boundaries of brain-computer interfaces (BCIs). Dynamic neural coding, inter-individual variability, low signal-to-noise ratio, partial observability and paradigm dependence jointly impose upper limits on decoding accuracy, information transfer rate, complex intention decoding, user experience and system stability. The authors propose information enhancement, adaptive decoding, human-machine collaboration and system optimization, arguing that gains will come from extracting more from the neural signal rather than from overcoming the underlying limits.

Large language models used to correct error-prone text from P300 brain-computer interface spellers can produce fluent sentences that no longer mean what the user intended, a failure the researchers call intent drift. Across 20 open-weight models tested on simulated P300 output, drift rose from 2.2% at a 0% character error rate to 60.3% at 40%, with an odds ratio of 2.30 per 10-percentage-point increase. Model confidence separated correct from incorrect outputs with an AUROC of 0.83 but was poorly calibrated: 28.4% of high-confidence outputs were unfaithful. Clinically important messages such as medication doses drifted slightly more than matched controls (odds ratio 1.10). None of six correction strategies eliminated drift; conservative strategies reduced it, but even the best still produced wrong output 18 times per 100 corrections. The study is a preprint and has not been peer reviewed.

A workshop at the BCI Society Meeting 2025, run with the Implantable BCI Collaborative Community (iBCI-CC), took up how clinical outcome assessments (COAs) for pivotal BCI trials should be selected, developed and validated. Participants pointed to patient heterogeneity, the absence of widely validated COAs, and the difficulty of capturing outcomes that matter in home and daily-life settings. The discussion lays groundwork for the iBCI-CC Clinical Study Endpoints Workgroup to build a transparent process for identifying meaningful aspects of health and concepts of interest, in support of regulatory approval and reimbursement.

EEG signals are only microvolts strong, so blinks, jaw clenching and muscle activity easily contaminate them. Researchers at Tianjin University and Tiangong University in northern China proposed MTGNet, a framework that suppresses these artifacts while preserving the information downstream tasks need. On the public EEGDenoiseNet dataset, it cut spectral relative root-mean-square error by 18.9%, 31.5% and 14.0% for EMG, EOG and mixed artifacts respectively; on a real-world fatigue EEG dataset, it raised classification accuracy by 6.20 to 6.69 percentage points over unprocessed input. Adapting the framework to a new task takes only 0.33 million low-rank adaptation (LoRA) parameters and no paired clean EEG reference.

LibriBrain100, a magnetoencephalography (MEG) dataset released on August 25, 2026, contains more than 100 hours of high-quality recordings, about 80 of them from a single subject, the deepest within-subject collection of its kind. It is meant as a standardized benchmark for neural speech decoding and ships with open-source tooling and an online competition; a further 32 subjects contribute roughly 40 minutes each to offset thin per-subject data. Using an existing decoding model, the team reported state-of-the-art results on a word-classification benchmark, which it takes as evidence of both data quality and the value of deep within-subject recording.

A preprint introduces the von Neumann State-Space Transformer (VN-SST), a neural decoding model that is more data-efficient than a modern Transformer across three motor-cortex decoding benchmarks, with the largest gains where training data is scarce.