24 articles
A KU Leuven team led by Peter Janssen trained three rhesus macaques to pilot avatars through virtual environments using implanted multi-array BCIs, with premotor cortex outperforming primary motor cortex and a single fixed decoder generalising across perspectives, avatars, and dynamic targets.
A Mark Hersam-led team at Northwestern has demonstrated aerosol-jet-printed neuromorphic devices that generate biologically realistic voltage spikes and trigger responses in mouse cerebellum slices, a first for printed electronics and a potential route to lower-cost, flexible neural interfaces.
The Barcelona neurotechnology enrolled ten patients and implanted eight during brain tumour resection at Salford Royal Hospital, completing the first human cohort for any graphene-based neural interface. Full results due later in 2026.
A UC Irvine-led team has demonstrated the first brain-computer interface that simultaneously controls a walking exoskeleton through decoded brain signals and feeds artificial leg sensation back to the user's sensory cortex — closing the loop on locomotion for the first time.
A Purdue team reports in Science that a liquid monomer injected into the motor cortex self-assembles into a conductive electrode mesh — catalysed by the iron in blood — and can then modulate neural activity with external near-infrared light.
A new 42-month programme will fund teams to build biological processing units from neural organoids, benchmarked against video games and drone navigation, at energy budgets rivalling a fruit fly's brain.
A Johns Hopkins team reports 77.5% speech classification and 78–84% motor decoding accuracy using Precision's 1,024-channel thin-film cortical array during awake brain surgery in four patients.
Tohoku University researchers trained cultured rat cortical neurons to produce sine waves, square waves, and chaotic Lorenz attractors using closed-loop FORCE learning, demonstrating that biological neural networks can serve as computational resources.
University of Iowa researchers have demonstrated for the first time that non-invasive transcranial magnetic stimulation can modulate hippocampal activity when the stimulation site is tailored to each patient's brain connectivity.
A Brown University-led trial published in Nature Biomedical Engineering demonstrates simultaneous motor and sensory restoration in three people with complete spinal cord injuries using patterned electrical stimulation and machine learning.
A UMC Utrecht study found that a brain-computer interface trained on daytime neural signals produced 245 false commands per hour when the patient slept, exposing a fundamental problem for 24/7 assistive devices.
A new hafnium oxide memristor from the University of Cambridge mimics biological synapses with a million-fold reduction in switching current, pointing toward more efficient neuromorphic hardware.
A brain-computer interface developed by Mass General Brigham and Brown University enabled two paralysed individuals to type at speeds rivalling able-bodied texting, with a word error rate of just 1.6 per cent.
New research upends a foundational assumption in neuroscience and BCI design: that individual frequency bands reliably map to neural activity. The discovery could reshape how we decode brain signals.
Researchers developed a near-invisible BCI technique using texture-based visual stimuli to monitor mental workload in pilots without the distraction of traditional flickering screens.
A new iontronic reservoir built from self-healing hydrogel creates neuromorphic prostheses that can withstand the unpredictable physiological environment inside the human body.
A researcher has compiled an index of datasets used to reconstruct what people see from brain activity, potentially accelerating development of visual BCIs and neuroprosthetics.
A comprehensive review maps how brain signals measured during stroke recovery are converging with BCI systems and closed-loop neuromodulation to move rehabilitation beyond generic protocols toward individualized treatment.
New framework resolves the historic trade-off between acoustic richness and linguistic clarity in neural speech decoding, achieving 4.0/5.0 intelligibility with minimal training data.
Researchers developed TFSANet, a deep learning model that decodes language comprehension from EEG signals, achieving 60.73% accuracy in aphasia patients using Chinese linguistic stimuli.
New fMRI research identifies precise cortical locations for implantable BCI electrodes, finding that surface recordings from gyral regions may suffice for gesture decoding without penetrating sulcal depths.
Researchers adapted automated fuzzing from software security to stress-test machine learning models that control neural implants, revealing widespread violations of biophysical safety limits in retinal and cortical devices.
New research reveals how targeting different brain regions with theta burst stimulation enhances motor imagery performance through separate mechanisms, offering potential solutions for stroke patients struggling with BCI control.
Researchers demonstrate that individualized fMRI models can decode spontaneous pain in chronic pain patients, marking a shift toward objective, brain-based pain measurement that could reshape treatment and neurotech interventions.
Get the latest BCI news, breakthroughs, and market moves delivered to your inbox.
No spam. One-click unsubscribe. We respect your privacy.