Research news

Marom Bikson and colleagues tested whether people with major depression could tolerate a stronger, faster course of transcranial direct current stimulation, which passes a weak electrical current through electrodes on the head. A home-based 2 mA version was recently approved by the US Food and Drug Administration for depression, but skin discomfort has limited higher currents. Using a high-capacity stimulator and hydrogel electrodes, the team gave 20 patients 20-minute sessions at 6 mA, twice a day for five days. All 203 sessions were completed, and 19 of 20 people tolerated the full dose. There were no serious adverse events. Everyone had skin redness at the electrodes, with no severe reactions. Depression scores fell by more than half, though the study had no placebo comparison. The researchers say a trial comparing it with a sham treatment is warranted. The study, posted on medRxiv, has not yet been peer reviewed.

Students increasingly turn to online videos for class help. Lucas Parra and colleagues asked whether that habit could help learning in a college course. In an introductory physics class, 103 undergraduates were assigned to one of two groups. One group did conventional reading assignments and exercises. The other searched online for instructional videos of their own choosing and answered "association questions" meant to guide their engagement with the material. The team also recorded working memory, motivation, and GPA.

The video group showed evidence of better learning on quizzes and on physics knowledge tests given before and after instruction. On exams the evidence was indirect, since performance depended on how much students engaged with the questions. Working memory, GPA, and motivation predicted most outcomes. The authors conclude that pairing video searches with guiding questions improves learning by prompting active engagement. The study appears in the Journal of Science Education and Technology.

Transcranial focused ultrasound uses sound waves to change activity in deep brain regions without surgery, but its effects vary and are not fully understood. Jacek Dmochowski and a collaborator turned to stochastic thermodynamics, a branch of physics concerned with how irreversible a process is, meaning how different it would look played backward in time. The measure is called the entropy production rate.

Reanalyzing published recordings from freely moving mice receiving ultrasound to the thalamus, they found that a light-based signal of neural activity rose steadily with dose, while the entropy production rate peaked at moderate intensity. Right after stimulation, it stayed elevated but was not tied to the activity signal's size. Trials starting with lower irreversibility changed more during stimulation. The results suggest ultrasound reshapes brain dynamics beyond what response strength captures. The study, posted in eLife, is a preprint that has not yet been peer reviewed.

Inflammatory signaling molecules such as IL-1, IL-6, and NF-κB are known to drive disease in specific organs, but they are usually studied without regard to the seasons. In a paper in the journal Biology, Bingmei Fu of CCNY and colleagues compared this research with the Yellow Emperor's Inner Canon, a classical text of traditional Chinese medicine.

Using an existing framework that links concepts in the text, such as Cold and Dampness, to specific molecules, the team translated the text's descriptions of an inflammatory condition called Bi syndrome into modern terms. They then searched published studies in the PubMed database for supporting evidence. The authors report that the literature fits a seasonal pattern described in the text, with bone affected in winter, tendons in spring, blood vessels in summer, muscle in late summer, and skin in autumn. They propose this as a unified framework for understanding seasonal inflammatory disease.

Measuring how stiff a bacterial cell is usually requires specialized tools such as atomic force microscopes or microfluidic devices. In a preprint on bioRxiv, which has not yet been peer reviewed, Kelsey DeFrates and colleagues describe a simpler method called mechanofiltration that uses standard lab equipment. Bacteria are spun in a centrifuge through filters with pores narrower than the cells, so the cells must squeeze to pass. Combining counts of cells that get through with measurements of cell size gives an estimate of how much they can deform, a stand-in for stiffness.

In E. coli, the test detected known softening caused by genetic changes to the cell's outer layers, membrane-disrupting chemicals, and antibiotic doses too low to kill. Results closely matched earlier measurements made with other methods, and the test reproduced known differences between species. The team describes it as an inexpensive, scalable way to screen many samples.

Artificial neural networks are often called black boxes because it is hard to see how they work inside. In a preprint posted to arXiv, which has not yet been peer reviewed, Lucas Parra and colleagues report that training these networks produces local symmetries that mathematicians call fibrations and coverings. The team proved that a widely used training method, stochastic gradient descent, tends to settle into covering symmetries and stay there. They observed these symmetries in several major network designs, including multilayer, convolutional, recurrent, and transformer networks.

The researchers used the symmetries to shrink networks to 17 percent of their original size without losing performance. Breaking the symmetry in a controlled way also helped networks keep their ability to learn new tasks over time, with results the authors describe as state of the art. Parra and colleagues suggest the approach could make AI systems easier to interpret and more efficient.

Jacek Dmochowski and colleagues studied transcranial photobiomodulation, a technique that shines near-infrared light through the head to stimulate mitochondria, the energy-producing parts of cells. The approach may have antidepressant effects. In work published in Brain Stimulation, the team delivered the light to both sides of the prefrontal cortex, at the front of the brain, in people with major depressive disorder while recording brain scans with functional MRI. They compared low, medium, and high (pulsed) doses with a sham, or fake, treatment.

A medium dose increased the strength of the blood-oxygen signal measured by the scanner, while a low dose decreased it. The high dose and sham produced no significant change. The effects reached beyond the areas directly lit. A single session did not change depression severity compared with sham, and all doses were well tolerated. The researchers say the results may help guide dose choices in future studies.

Transcutaneous auricular vagus nerve stimulation (taVNS) sends mild electrical pulses through the skin of the ear to activate the vagus nerve, part of the body's rest and digest system. Marom Bikson and colleagues tested how electrode location and pulse pattern affect heart and breathing. In a randomized study, 25 healthy participants each received four stimulation settings at the tragus, the small flap in front of the ear canal, and the same settings at the earlobe for comparison. That made 200 sessions in total. Heart rate, heart rate variability, and breathing were recorded throughout.

Only one setting, 25 pulses per second on the left ear alone, raised heart rate variability compared with earlobe stimulation. This is a sign of increased vagal activity. Averaged over all stimulation periods, it was also linked to higher heart rate. Breathing did not change. The other settings showed no clear differences. The study, posted on bioRxiv, has not yet been peer reviewed.

Brain researchers often use a standard "average" brain, the MNI152 template, as a common reference in imaging studies, computer models, and brain stimulation research. Marom Bikson of CCNY's Department of Biomedical Engineering and colleagues tested how well it reflects real brains. In Biomedical Physics & Engineering Express, they compared MNI152 with 430 MRI scans of Asian, Black, and White participants from a public dataset, and with a newer template called US200.

The team measured how much each template had to stretch or shrink to fit each scan. MNI152 consistently had to be shrunk overall, and its shape mismatches varied from region to region. Simple linear alignment alone could not remove these differences. For this group of scans, US200 matched real anatomy more closely. The authors say relying on MNI152 may introduce anatomical bias where precise brain shape matters. They support building modern templates, possibly for specific groups.