Text-message tutoring keeps kids learning maths when schools close

Hello Nature readers,
Today we discover that Denisovans made excellent deer hunters, hear that texts and phone calls can keep kids learning when schools are forced to shut and learn that Beatles puns are here, there and everywhere in scientific literature.

If you enjoy this newsletter, please consider recommending it to a friend or colleague. Click here to forward it by e-mail. Thank you!

Fossil trove sheds new light on Denisovans

A collection of fossils and artefacts recovered from an abandoned quarry in southwestern China has given researchers an unprecedented glimpse into the lives of ancient humans called Denisovans. The items suggest that Denisovans — who lived at the site for at least 30,000 years — crafted bone tools similar to those of the Neanderthals, and were skilled hunters of animals such as deer. “This site’s the full package,” says archaeologist Ben Marwick. “It’s the first site that we have with a long, rich sequence that’s uniquely associated with Denisovans.”

Texts and calls keep learning afloat

Text messages and weekly 20-minute phone calls from a teacher can improve children’s learning during prolonged school closures. The approach offers a relatively cheap way to support children’s mathematics education when schooling is disrupted by natural disasters, pandemics or other emergencies, say the study authors. In a trial across five countries, kids that received both texts and phone calls improved the most. But text messages alone also improved children's learning compared to a control group.

Virtual cell model predicts best cancer drug

A prototype tool called ProteinTalks can predict how the cells in triple-negative breast cancer (TNBC) tumours — which can behave very differently at the molecular level — will respond to various drugs. Researchers trained the computational simulation on millions of protein measurements collected over time from breast cancer cell lines that were treated with dozens of antitumour drugs and drug combinations. The model did a good job of predicting how cells would respond to drugs it hadn’t seen before and real-world clinical outcomes.

The mathematician’s protégé

In 2019, retired topologist Joan Birman received an email from 15-year-old Vasudha Bharathram asking for help in learning geometry. After a few meetings, Birman realized that Bharathram had a knack for the subject, and often solved well-known problems in an unconventional way. Bharathram inspired Birman to revisit a problem that had eluded her for decades: is the Burau representation — the translation of the structure of braids into algebraic objects called matrices — faithful for four-strand braids? This year, a 99-year-old Birman, Bharathram and collaborator Tara Brendle published their answer: yes.

AI is a debate partner, not an oracle

The concerns that researchers might outsource scientific judgement to generative artificial intelligence are valid, but aimed at the wrong object, argues computer scientist Marc Schmitt. The danger is not AI, but people’s assumption that chatbots are producing final answers. Users should treat AI tools as intellectual sparring partners, with which they can develop and hone their thinking. AI developers should encourage this argument-based dialogue with chatbots, and universities should treat a student’s ability to participate in these arguments as a core skill, Schmitt suggests.

Video: bacteria shuttle drugs to tumours

To minimize the off-target effects of cancer therapies, researchers need to find ways to deliver drugs to tumours accurately. Some species of bacteria are attracted to the low-oxygen environment in tumours, which makes them ideal candidates to shuttle drugs directly to their target. These bacteria can also be engineered to manufacture the drug inside their cells, and even to sense the conditions of their environment to ensure that they only release their cargo inside a tumour.