Every new paper, in one sentence you can actually read.

Each issue takes 15 brand-new scholarly papers, says in a single sentence what each one found, then explains it. Every paper is screened for retractions and a live DOI before it reaches you.

This screen checks process signals. It does not tell you the findings are true.

Get it free, every two days

Five papers free every issue, about 15 emails a month. No spam. Leave in one click.

Prefer no email? The same issues are in the RSS feed.

In this issue 15 papers · 5 free

01 Proceedings of the National Academy of Sciences Barry Bloom and the convergence of immunology, infectious disease, and public health Free
02 PNAS Nexus Soluble Siglec-9 combined with exercise ameliorates HFD-induced metabolic and depressive-like phenotypes with gut-brain axis changes Free
03 Nature Communications Lung environment in healthy old age shapes the phenotype and CCR2-mediated recruitment of a subset of apoptotic, high-turnover alveolar macrophages Free
04 Nature Retrofitting language models to operate over bytes Free
05 Nature Communications Reward function compression facilitates goal-dependent reinforcement learning Free
06 Proceedings of the National Academy of Sciences Bone morphogenetic protein signaling as a key regulator of retinal fibrosis Subscribers
07 Proceedings of the National Academy of Sciences Infection of macrophages by mpox virus activates inflammasome and cell death, informing multitarget therapeutic strategies Subscribers
08 Nature Ecology & Evolution Biodiversity safeguards ecosystem services and functions worldwide Subscribers
09 The Astrophysical Journal Local Heavily Obscured Active Galaxies Observed with Chandra. I. Spectral Properties of the Nuclear and Extended X-Ray Emission Subscribers
10 PNAS Nexus Scientific machine learning for modeling complex fluids with thixotropy and plasticity Subscribers
11 Nature A nuclear clock synchronized to 229Th Subscribers
12 Nature 5-oxoETE links redox control of epithelial damage detection and resilience Subscribers
13 Physical Review X Information Thermodynamics of Agents: The Work Capacity of Channels with Memory Subscribers
14 Nature Communications Post-fabrication reconfiguration of hydrogels through thermodynamic asymmetry in dynamic chemistry Subscribers
15 Geophysical Research Letters Hydrological Extremes Increase Water‐Resource Constraints Under Climate Change Subscribers

The papers Read free

Proceedings of the National Academy of Sciences·7 October 2026

Sarah M. Fortune, William R. Jacobs et al.

A tribute honors immunologist Barry Bloom, who shifted immune research toward real disease germs.

Immunologist Barry Bloom died in early 2026. He helped change how scientists study the immune system, moving focus from simple lab molecules to real disease germs like the one that causes leprosy. He also led a major public health school and trained many scientists who still carry on his work.

What it could change A career that helped move immune research from simple molecules to real disease germs.

This describes a scientist's legacy, not a new experimental result.

A tribute essay, not a new study with data.

Integrity screen: passed (3 checks) Checked 8 October 2026. Sarah M. Fortune, William R. Jacobs et al., Barry Bloom and the convergence of immunology, infectious disease, and public health Read the source

PNAS Nexus·6 October 2026

Kohki Matsubara, Shintaro Tsukada et al.

In mice, a protein plus exercise reduced depression-like behavior caused by a fatty diet.

In mice, a natural anti-inflammatory protein combined with exercise eased depression-like symptoms caused by a fatty diet. The protein calmed swollen brain cells and helped new brain cells grow in a memory area. It also shifted gut bacteria toward healthier types that produce helpful byproducts. Mice fed the diet for 8 weeks, then treated for 4 more weeks, showed the biggest improvement when the protein and exercise were used together.

What it could change A protein and exercise together eased diet-linked depression symptoms in mice more than either alone.

Based on mouse brain, gut bacteria, and behavior tests, not human trials.

Mice only, four week treatment, no human testing yet.

Integrity screen: passed (3 checks) Checked 20 October 2026. Kohki Matsubara, Shintaro Tsukada et al., Soluble Siglec-9 combined with exercise ameliorates HFD-induced metabolic and depressive-like phenotypes with gut-brain axis changes Read the source

Nature Communications·7 October 2026

Susanta Pahari, Miranda Lumbreras et al.

In mice, monkeys, and humans, one lung immune cell type grows over 10 times more common with healthy aging.

A certain lung defense cell becomes far more common as mammals age in healthy ways. It grows to more than 10 times its normal level in old mice, monkeys, and people. Old lungs push these cells to change shape and bring in new short-lived replacement cells. These new cells die quickly through a natural cell death process, which may add to age-related inflammation.

The number more than 10-fold increase

The rise in one lung immune cell type seen in old mice, monkeys, and humans.

Mouse mechanism work only, human data limited to cell counts.

Integrity screen: passed (3 checks) Checked 8 October 2026. Susanta Pahari, Miranda Lumbreras et al., Lung environment in healthy old age shapes the phenotype and CCR2-mediated recruitment of a subset of apoptotic, high-turnover alveolar macrophages Read the source

Nature·7 October 2026

Benjamin Minixhofer, Tyler Murray et al.

Converted letter-by-letter AI models now match word-chunk models in performance.

AI language models usually break text into word chunks before reading it. This hides small details, which causes problems when every letter matters, like in computer code or biological sequences. A new method changes existing word-chunk models into models that read text one letter at a time. These new models catch up to the older ones and do better on tasks that need letter-by-letter attention.

What it could change A way to turn word-chunk AI models into letter-reading models without full retraining.

Shows letter-level models can match word-chunk models while reading text more precisely.

Method paper, no reported test scores or sample sizes yet.

Integrity screen: passed (3 checks) Checked 20 October 2026. Benjamin Minixhofer, Tyler Murray et al., Retrofitting language models to operate over bytes Read the source

Nature Communications·8 October 2026

Gaia Molinaro, Anne G. E. Collins

Learning speeds up when people can simplify many goals into one simple rule.

People learn faster when their goals can be boiled down to a simple rule. In six experiments, people learned worse when they had many different goals to juggle. They learned better when those goals followed one simple pattern. Faster mental processing of rewards also matched better learning, and computer models linked this speed to more accurate choices.

What it could change People learn new goals faster when they can shrink many goals into one simple mental rule.

This describes a pattern seen across six behavioral experiments, not a brain mechanism directly observed.

Six lab experiments on humans, mechanism not directly confirmed yet.

Integrity screen: passed (3 checks) Checked 8 October 2026. Gaia Molinaro, Anne G. E. Collins, Reward function compression facilitates goal-dependent reinforcement learning Read the source

10 more papers are in the full issue.

The same screening, the full reading, every two days. 2 euros a month, or 20 a year, and you can leave in one click.

Proceedings of the National Academy of Sciences·7 October 2026

Sonali Sharma, Gieth Alahdab et al.