Brain & Mind Weekly — Mar 09, 2026

A drug that teaches a broken gene to work harder just slashed seizures by up to 91% over the first 20 months in children with one of epilepsy's cruelest forms — and it's the first gene-regulation therapy to reach this stage for any epilepsy. Meanwhile, anatomists found a brain barrier nobody knew existed, addiction researchers traced cocaine relapse to a single molecular bottleneck, and a cluster of lab-grown neurons learned to balance a virtual pole. The brain, it turns out, is still full of surprises — and so are the tools we're building to study it.

Dravet syndrome is a genetic epilepsy that begins in infancy and doesn't relent. Children average seventeen seizures a month. The cause is deceptively simple: one copy of a gene called SCN1A doesn't produce enough of a protein that nerve cells need to fire properly. Every existing treatment manages symptoms without touching the underlying deficiency.

This week, the New England Journal of Medicine published results from 81 children treated with zorevunersen, a drug that takes a fundamentally different approach. It's an antisense oligonucleotide — a short, engineered strand of genetic material that intercepts the cell's RNA machinery and coaxes the healthy copy of SCN1A to produce more protein. It doesn't edit DNA. It doesn't delete the faulty gene. It amplifies the working one.

The numbers were striking: children on the highest doses saw convulsive seizures drop by 59% to 91% over the first 20 months. Crucially, improvements extended beyond seizure counts to language, motor skills, and behavior — suggesting the drug isn't just suppressing electrical storms but allowing developing brains to catch up. An accompanying editorial called it a potential watershed for genetic neurology.

A global Phase 3 trial (EMPEROR) is now enrolling roughly 150 patients, with results expected by 2027. If those results hold, this becomes the first approved disease-modifying treatment for any genetic epilepsy. For families who've had nothing but symptom management, that's not incremental. It's everything.

The hardest part of cocaine addiction isn't stopping — it's staying stopped. Cravings can surge months or years later, triggered by a place, a face, a smell. This week, researchers at Michigan State University published work in Science Advances that traced that relapse biology to a specific molecular address.

The team focused on the circuit connecting the ventral hippocampus (the part most involved in emotional memory and context) to the nucleus accumbens (the brain's reward hub). They found that repeated cocaine exposure doesn't just flood the reward system temporarily — it changes which genes are switched on in that circuit, reshaping how excitable those neurons become. The key player is a transcriptional regulator, a protein that sits inside the cell nucleus and controls which genes get turned on or off. When cocaine history alters this protein's activity, the result is lasting biological changes that prime the animal to seek cocaine again when familiar cues reappear.

This is rodent data, and the translational gap to humans is real. But the importance is the level of specificity: not "the reward circuit is involved in addiction" — we've known that for decades — but which cell type, in which projection, controlled by which molecule, drives the specific act of relapse. That precision is what makes a drug target.

Here is a sentence that should probably not be true in 2026: scientists just discovered a new brain barrier. Not a better description of a known one — an entirely new structure.

Published this week in Nature Neuroscience, Verhaege and colleagues describe an overlooked barrier at the base of the choroid plexus, a structure tucked inside the brain's fluid-filled ventricles that produces cerebrospinal fluid. The blood-brain barrier was discovered in the 19th century, and additional barriers have been catalogued since. But this one — adding another layer of selective control over what moves between the brain's fluid compartments — was hiding in plain sight.

Think of it like discovering a previously unknown security checkpoint in a building where you thought you already knew every exit and entrance. Why does it matter clinically? It matters amid longstanding difficulties delivering drugs into the brain, since every unknown checkpoint is a potential reason promising drugs fail to reach their targets. The choroid plexus is also implicated in Alzheimer's disease, aging, and neuroinflammation. Expect a surge of follow-up work asking what this barrier lets through — and what it doesn't.

How do your fingertips distinguish silk from sandpaper, or a breeze from a poke? For decades, the precise wiring was a mystery. This week, David Ginty of Harvard and Patrik Ernfors of the Karolinska Institutet won the 2026 Brain Prize — the world's largest neuroscience award — for cracking it.

Their work identified the different types of sensory neurons in skin responsible for detecting specific sensations (pressure, vibration, temperature), then traced the pathways those signals take from the body, up the spinal cord, and into the brain. The result is essentially a blueprint: a map of normal touch that also pinpoints where things go wrong in chronic pain and may help explain the sensory sensitivities often experienced in autism. When you can name the parts, you can start fixing them.

When you're trying to figure out what someone else is thinking, your brain isn't running one fixed empathy program — it's adjusting its strategy on the fly. A new Nature Neuroscience paper used a multiplayer game plus computational modeling to show that the brain carries a specific neural signature for how flexibly we update our models of other people's minds.

Volunteers played a game where the other player could change strategies over time. The researchers built a model (called CHASE) that tracks how people update beliefs about the other person's reasoning level, then aligned those trial-by-trial estimates with brain activity. The classic theory-of-mind regions — medial prefrontal cortex, temporo-parietal junction, precuneus — weren't just on or off. Their activity patterns tracked how participants shifted between mentalizing strategies, not just the content of specific beliefs. People whose brain activity more closely tracked the computational measure were better at predicting their partner's future moves — suggesting a parametric neural marker of social skill.

This model-based approach matters because it treats social cognition like a computational problem with measurable parameters, not a vague "empathy region lights up" story. If it replicates in clinical populations, it could become a biomarker for social-cognitive disorders.

3D Electronic Mesh for Organoid Recording — Northwestern University's soft, porous 3D electronic mesh, published in Nature Biomedical Engineering, is now available as a described protocol for labs wanting whole-organoid electrical mapping. The "pop-up book" design transforms from a flat lattice into a 3D scaffold that cradles a mini-brain while staying porous enough for nutrient exchange. This is the step that turns organoids from observation platforms into genuine experimental systems for drug screening and disease modeling.

Wireless Cross-Species EEG System — A new open-access, non-invasive wireless EEG system demonstrated reliable sleep-wake tracking across reptiles, birds, and mammals without cranial implants or tethering. Designed for comparative neuroscience labs, it meets modern animal welfare standards while producing ecologically valid data. Complementary reports describe a wearable human EEG device with hospital-grade accuracy for under $500 — if validation holds, it could democratize brain monitoring.

Bicistronic Dual-Indicator Viral Vector — A new viral construct co-expresses voltage and calcium indicators in the same neurons, enabling simultaneous recording of membrane potential and intracellular calcium in awake, behaving mice. Early results show that subthreshold depolarizations — voltage changes too small to trigger a spike — can still drive strong calcium signals, meaning the space of inputs that can induce synaptic plasticity may be much larger than assumed.