iiinova — where your digital self keeps evolving

This study was led by Alex Utzschmid and @jonasterlau.bsky.social. All of our BCI work is the result of a great interdisciplinary, collaborative effort. Thanks to Bernhard Meyer, Arthur Wagner, Jens Gempt, Lisa Held, @lauraschiffl.bsky.social @86chenhongbiao.bsky.social Göktug Alkan, Paolo Favero

020

Short- and long-term neuronal stability was comparable to previous findings in BCI participants without structural brain lesions. Together, the results show that major structural brain injury does not preclude stable intracortical recordings.

Across 111 recording sessions over 1,240 days, unit yield and signal quality remained stable. Individual neurons could be tracked across sessions - even after extended recording gaps. Tracked neurons also retained consistent spiking properties.

We studied a participant with chronic language impairments (aphasia) after a large left-hemispheric stroke. Four 64-channel microelectrode arrays were implanted in the right hemisphere, targeting regions corresponding to the damaged language network.

Why this matters: BCIs are mainly being developed for spinal cord injury and motor neuron disease. But stroke is the most common cause of acquired brain injury and a leading cause of long-term disability - potentially one of the largest future application fields for BCIs.

220

Jacob Lab @jacob-lab.bsky.social · 31/07/2026

A #stroke can reshape the brain. Can a #brain-computer #interface still reliably record single neurons from an injured brain for years? Our preprint provides the first evidence that it can - with stable recordings spanning 1,240 days! www.biorxiv.org/content/10.6... #BCI #neuroskyence 🧠⚡🧪 🧵👇

1277

🚨 New preprint 🚨 "Closed-loop sensory feedback enables fast and reliable instrumental acquisition in head-fixed mice" Giving mice the chance to self-correct their choices cuts the training time in half! www.biorxiv.org/content/10.6... #neuroskyence #ReinforcementLearning #CognitiveFlexibility 🧪

0165

8/8 Our findings point to a solution to the stability–flexibility problem: learning may establish a stable temporal scaffold while leaving neuronal selectivity free to adapt.

051

7/8 We developed Sparse Tensor Component Analysis to connect these population and single-neuron results. STCA suggests that dynamic selectivity can arise through rule-dependent recombination of stable task representations, with neurons changing how strongly they participate across sessions.

6/8 At the population level, choice representations generalized across rules. What changed was how the currently relevant cue representation aligned with the choice representation as each new rule was learned.

5/8 What the neurons responded to behaved very differently. Neurons repeatedly gained, lost, or changed selectivity - even after their activity profiles had stabilized. Stable activity therefore did not imply a fixed functional role.

4/8 Learning progressively stabilized when individual neurons were active during a trial. Once established, these temporal activity profiles persisted across later learning stages and rule switches.

3/8 We tracked the same medial prefrontal neurons over months while mice learned three successive cue–action–outcome rules. The cue–action relation changed, but the sequence and timing of trial events stayed constant, allowing us to separate activity timing from task selectivity.

2/8 We often infer a neuron’s function from two properties: when it is active and what it responds to. These are usually treated as one. But they need not change together - and separating them may reveal what learning actually stabilizes.

🚨 New preprint 🚨 How does the brain learn new things without rebuilding what it already knows? www.biorxiv.org/content/10.6... Fantastic work led by @yuyanghuang.bsky.social and Leonie Mehrke, with Tobias Bernklau and @laurabusse.bsky.social #neuroskyence #neuroscience #learning #PFC 🧪🧠 🧵👇

15114

Researchers with strong backgrounds in machine learning and AI, and a passion for advancing implantable neurotechnology, are also encouraged to apply. 📆 Application deadline: 15 Sept 2026 Please repost to help us reach the best candidates!

021

We’re especially interested in recruiting for: • implantable systems & hardware integration • embedded AI & edge inference • closed-loop stimulation • hybrid neurotechnologies

#Munich has become one of Europe’s leading hubs for implantable #neurotechnology, with recent milestones including • the first intracortical microelectrode BCI for people with severe language disorders • Europe’s first intracortical BCI enabling restoration of mobility after spinal cord injury

TUM - Tenure Track Assistant Professor in » Implantable Brain-Computer Interfaces «

🚨 We’re hiring! 🚨 The Technical University of Munich @tum.de is recruiting a Tenure-Track Assistant Professor in Implantable Brain–Computer Interfaces! We’re looking for someone to help build the next generation of robust, clinically scalable #implantable #BCIs. portal.mytum.de/jobs/profess... 🧵👇

portal.mytum.de

Studierenden- und Mitarbeiterportal der Technische Universität München

11817

Reposted by Jacob Lab

Technische Universität München @tum.de · 06/05/2026

A research project at our university investigates how #BrainSignals can be translated into digital commands, aiming to enable people with #paralysis to interact with computers or #RoboticSystems: go.tum.de/611249 #AI #Neuroscience 📷A. Eckert

go.tum.de

The power of the mind

Michael Mehringer is paralyzed from the neck down. Researchers at TUM are working to enable him to regain greater autonomy.

074

Jacob Lab @jacob-lab.bsky.social · 10/04/2026

Great work from @alexanderhuth.bsky.social and team! Widespread, bilateral #semantic representations in #aphasia decoded by fMRI. This aligns very well with our recent #single #unit findings: www.biorxiv.org/content/10.6... Language BCIs for aphasia are absolutely within reach! #neuroskyence 🧪

biorxiv.org

0122

Reposted by Jacob Lab

Technische Universität München @tum.de · 02/04/2026

Researchers at our university observed visual signal flow in a specific thalamocortical pathway, supporting key aspects of the 1981 Nobel Prize model by Hubel and Wiesel and refining the understanding of cortical processing: go.tum.de/450857 #Neuroscience #VisualPerception 📷A.Eckert

go.tum.de

Brain researchers provide evidence for a theory of visual perception

Researchers at TUM have mapped and demonstrated the process of visual perception at the cellular level in the brain.

052

Reposted by Jacob Lab

WilluhnLab @willuhnlab.bsky.social · 01/04/2026

Submit your abstract until May 15 for the "Monitoring Molecules in Neuroscience" conference in Amsterdam (Sept 7-10, 2026) to get the latest update on in-vivo neuro-methods from a brilliant lineup of keynote and panel speakers: monitoringmolecules.org/invited-spea... Please repost!

monitoringmolecules.org

MMiN 2026 - MMiN

045

A special part of this long journey has been the close collaboration with the Department of #Neurosurgery @tum.de (Jens Gempt, Arthur Wagner, Bernhard Meyer). This project would not have been possible without that partnership, bringing together clinical insight, technical expertise and a common goal

010

In short, this is evidence for well-structured, organized language processing in the right hemisphere, with considerable symmetry to the left. Our findings also open new directions for developing #neurorehabilitation strategies and #BCI #technology for people who are lost for words.

Single-neuron activity and population dynamics were predictable using large language model (LLM) embeddings. Semantic features preferentially accounted for prefrontal activity, and phonological features for parietal activity.

130

Right-hemispheric frontoparietal neurons robustly participated in comprehension, retrieval and articulation of words, the core operations of language. Distinct subpopulations encoded lexical, word-level information in a highly task-specific manner, despite correlated firing across tasks.

100

Led by @lauraschiffl.bsky.social and Lisa Held, this study represents the largest dataset of its kind, with #microelectrode recordings from more than 10,000 neurons collected over ten months in an individual with #aphasia (chronic language impairments) following #stroke

101