auroras on Earth and defines the extent of the solar system, these stars' powerful winds blast material into space. Meanwhile, the most massive of these stars eventually explode as supernovas, pushing even more gas aside. The collective radiation from both these old and young stars carves the nebula's giant cavity.

So what happened to that expelled gas? Most of it has been pushed outward and compressed into a dense shell of gas and dust surrounding the superbubble. Within that shell, new stars are forming ‪—‬ so the same massive stars that destroy the gas cloud that created them also may help to produce their replacements.

This is known as "stellar feedback" — when massive stars heat, ionize and disperse the gas and dust around them, making star formation impossible in their vicinity. Meanwhile, their winds can squeeze neighboring gas clouds until they become dense enough to collapse under their own gravity and make new stars.

This Hubble image of N44 contains almost half a million stars, including those in the nebula and some lying along the same line of sight. About 30,000 of them are young stars yet to fuse hydrogen in their cores — the reaction that powers stars like the sun. With so many young protostars not yet hot and dense enough to become full-fledged main-sequence stars, N44 gives astronomers a rare opportunity to attempt an answer to what may seem like a very simple question: how long does it take to make a star?

'> Haunting Hubble telescope image captures cosmic cycle of destruction and creation —‬ Space photo of the week

But is that really the case? Are there hidden, more complex reasons for why cats lick each other?

While research into allogrooming is ongoing, scientists have suggested different explanations for why cats lick each other, including some social reasons and some more practical ones.

Social bonding or social tension?

The historical view, based on observations of free-roaming-cat colonies, was that allogrooming is a sign of affection and helps cats create and maintain bonds with other cats in their social group, perhaps by creating a shared group smell.

But a 1998 study of 25 cats living together in an indoor-outdoor enclosure poked holes in this theory. The study found that aggressive behavior showed up in about 35% of allogrooming sessions, which suggested that allogrooming might not be about friendship. Instead, the authors proposed, it may help to defuse a tense situation in a confined space, where an actual fight would be too risky.

To investigate, Morgane Van Belle, a cat behavior scientist at Ghent University in Belgium, and her colleagues analyzed videos of allogrooming from 53 two-cat homes. They concluded that allogrooming may have different social functions, depending on the situation.

The team noticed that the cats sometimes licked each other while huddling together and copying one another's body postures. In these moments, allogrooming probably helps strengthen social bonds, the researchers proposed. Most licking was directed toward the ears; this may help cats relax because the skin there is rich in sensory nerve endings, Van Belle said. Allogrooming produces a relaxing effect in other animals too. For example, it lowers the heart rate in horses and triggers the release of feel-good hormones in monkeys.

They also noted that allogrooming sometimes led to friendly wrestling, in which the cats grabbed each other with their front legs and tumbled around. This suggests that allogrooming may serve as a signal to start play.

In other situations, however, the researchers noticed that allogrooming was followed by behaviors linked to conflict, such as turning the ears backward, pawing, and biting. In these cases, one cat would stand or lean over the other, and both cats showed subtle signs of stress like lip licking, head shaking or self-scratching.

Based on these observations, Van Belle's team think that allogrooming may be an "appeasement signal" in fraught moments that could lead to conflict or injury, for example, when one cat wants to take over a sleeping spot already occupied by another. They think it possible that the licking helps to calm things down and keeps a spat from turning into a real fight. Similar behavior has been seen in meerkats, where subordinate individuals groom dominant ones to placate them.

Yet because this licking often targets the neck — the same area cats bite during fights — it also may signal a subtle aggressive threat, the researchers suggested.

Hygiene and grooming

Van Belle and her colleagues think that, beyond sending social signals, allogrooming may help with hygiene. After all, it's often aimed at the head and neck ‪—‬ the areas that are "least accessible for the cat itself during grooming," Van Belle said. Allogrooming is thought to have a hygienic function in other species too. For example, ants achieve colony-level hygiene by grooming heavily infected individuals more, and primates concentrate their grooming on body parts the recipient can't easily reach. When it comes to cats, however, more research is needed. Van Belle's study did not measure parasites, infections or skin cleanliness to see whether such grooming actually reduces dirt or bugs.

Dr. Terry Curtis, a veterinary behaviorist who has studied how often allogrooming happens, also thinks cats might groom each other for hygiene reasons. She wonders whether feral cats, who likely carry more fleas than house cats, might have a different allogrooming pattern.

The explanations proposed by Van Belle and colleagues are reasonable but not definitive, Curtis said. Because so many factors can shape a cat's behavior, she's skeptical that researchers will ever be able to say with real confidence exactly why one cat grooms another.

After working as a veterinary behaviorist for 25 years, Curtis has concluded it is often impossible to understand what's going on in a cat's mind.

"Congratulations, you got a cat," she said. "Let the mystery begin.”

How much of a cat fan are you? Find out by taking our cat quiz!

'> Why do cats lick each other?

It's odd to think of diamond, the hardest natural material on Earth, melting — but melt it does when blasted with extremely powerful lasers under the right conditions. Understanding how diamond responds to shock waves from lasers is an important part of developing nuclear fusion, the process that powers stars. Nuclear fusion is also a potential energy source for the future, so researchers have put a lot of effort into developing models that describe and predict how diamond behaves.

However, diamond is weird. Although the experimental data and theoretical models match up pretty well most of the time, there have been some strange discrepancies scientists haven't been able to explain. The biggest one is the 2,240 F (1,244 C) ‪—‬ roughly 20% ‪—‬ difference between previous experimental data and model-predicted melting temperatures of diamond. There's also been some debate about whether diamond reorganizes its atoms into a different kind of solid carbon before turning into a liquid at the end of the melting process.

Researchers have struggled to explain these discrepancies because the conditions diamond melts at are so extreme that it's extraordinarily difficult to measure it in labs on Earth. However, new experiments may finally offer the solution that scientists have pursued for two decades.

In a study published Aug. 13 in the journal Nature Physics, scientists zapped tiny plates of synthetic diamond with an ultraviolet laser, creating shock waves that were so powerful that as they passed through the samples, the diamond changed from transparent to mirror-like. The strong increase in reflectivity is one indication the diamond melted. By combining this change with measurements of how brightly the diamonds glowed while being zapped, the researchers mapped the melting temperature with great precision.

"We were able to take tiny diamond samples and shock compress them to temperatures hotter than the surface of the sun and to pressures higher than the center of Neptune and Uranus — and still measure atomic structure, temperature, density and optical reflectivity," study co-author Marius Millot, a research scientist at Lawrence Livermore National Laboratory in California, said in a statement.

The team found that the diamond sample's melting temperature was more than 1,300 F lower than previously thought — putting the melting point in line with theoretical predictions and finally explaining the long-held discrepancy.

The team also measured the samples' atomic structure with X-ray diffraction and saw that the diamond didn't transition to a different kind of solid carbon before melting, possibly because the energy required to rearrange the atoms was too large, the researchers wrote.

However, they also hypothesized that multiple shocks could be powerful enough for this transition to occur and that the way the shocks are applied to the diamond might affect how it changes phase. Understanding this is important for nuclear fusion research, as certain types of experiments involve lasers melting and crushing a diamond capsule to put the capsule’s contents, solid deuterium and tritium, under more than 30 petapascals of pressure and temperatures higher than 180 million F (100 million C), the requisite conditions for a fusion chain reaction to occur.

The researchers found that between about 660 and 1,060 gigapascals of pressure and at around 12,140 F (6,727 C), diamond exists as solid chunks floating in liquid carbon. As the pressure increases, more diamond transitions into liquid carbon, which is thought to be a very strange material. Unlike most forms carbon takes on Earth ‪—‬ like coal, graphite and diamond ‪—‬ liquid carbon is metallic, so it conducts electricity. It's also denser than diamond. So hypothetically, if you somehow were to put liquid carbon in a cup without instantly vaporizing it, a chunk of solid diamond could happily bob around in it like an ice cube in a glass of water.

Knowing how diamond behaves under such extreme conditions is also important for understanding the ice giant planets Uranus and Neptune. Based on measurements from the Voyager 2 spacecraft in the late 1980s and lab experiments on Earth, scientists think it literally rains huge chunks of diamond inside these planets and that their mantles may have liquid carbon oceans with diamonds floating around like icebergs. The new research means scientists can make better predictions about the planets’ interiors and their carbon cycles.

See how much you know about gemstones with our gold and gems quiz!

'> Scientists got diamond's melting point wrong by more than 1,000 degrees, crushing new laser experiment reveals

In the new study, published in July in the journal Nature, Huang and colleagues demonstrated that cancer-detecting AI might work better when it takes this humanized approach.

Training AI to hunt for cancer

AI algorithms called vision language models (VLMs) struggle with the first step that Huang described — that initial, cursory scan. That's in part because many pathology AI systems learn from what pathologists leave behind at the end of that search: a labeled image pointing out where the cancer is or an official diagnosis.

Instead, the researchers trained their new AI on pathologists' search behavior. They called this approach to training "Pathology-CoT," short for "chain of thought." It turns observable actions, including where pathologists move around and zoom in on an image, into training data.

To collect the data, the team created a tool that recorded how pathologists moved around a slide and changed magnification. The raw logs, gathered from eight pathologists, were messy, as a given pathologist might drift across a slide, overshoot their intended region of focus or fiddle with magnification to adjust it to their liking.

To clean up the data, the researchers filtered out those incidental movements, focusing on moments that appeared to represent deliberate attention, such as lingering over one view or making a sustained pan. Then, they compared those regions with eye-tracking data to check that the software was capturing where pathologists were actually looking.

For each region a pathologist inspected, the VLM also drafted a short rationale explaining why the region was worth examining and what features were visible; human pathologists could then accept, edit, or reject the rationale, creating additional training data for the AI. In one example, the AI flagged a portion of a slide as potentially metastatic and suggested zooming in to look for atypical cells. Other inspected regions were flagged as healthy tissue.

Ultimately, the researchers used this training method to build a new tool called Pathology-o3. It scans a slide at low resolution, uses a model trained on pathologists' behavior to choose regions worth a closer look, then sends higher-resolution views of those regions to a VLM for analysis.

Putting it to the test

Huang said the goal of the new study was not to show that Pathology-o3 worked better than specialized AI models that are specifically built to detect specific types of cancer; those models are often trained disease by disease. Rather, the researchers wanted to see whether their new training approach could help a general-purpose AI navigate a pathology slide more effectively.

They compared Pathology-o3 to other general-use AI systems, such as OpenAI's o3, and asked the algorithms to examine slides containing lymph node tissue. These slides were collected from colorectal cancer cases and some contained metastatic cancer, which human pathologists had already labeled.

The algorithm correctly identified slides that were positive for cancer 100% of the time. However, of the slides it identified as positive, 15.5% were actually negative. By comparison, OpenAI o3 correctly identified slides that were positive for cancer 87.5% of the time. Of the slides it identified as positive, 53.3% were actually negative.

The researchers designed Pathology-o3 to err on the side of flagging something for another look, rather than potentially missing cancer. That might help to explain the rate of false positives, Huang said.

Whether that rate of false alarms is acceptable depends on how Pathology-o3 is used, said Mohammad Asadi, a data scientist at Stanford University who was not involved in the research. It's not precise enough for the AI to diagnose patients on its own, but it could still be useful for a system to point a human toward regions of a slide that are worth double-checking. It may be an advantage that the approach shows the pathologist a specific region to inspect rather than declaring an entire slide suspicious, he said.

The researchers tried repeating the test on an independent dataset that the algorithms hadn't seen before to see how well it worked on unfamiliar slides. Pathology-o3 correctly identified slides that were positive for cancer 97.6% of the time. Of the slides it identified as positive, 37.1% were actually negative. The finding is an example of how AI performance can change when the data source changes, even when the medical task stays the same.

Asadi explained this result suggests the system can still work with slides from a different source. But that result does not yet show that using this tool would make pathologists more accurate or efficient in practice.

Can it help pathologists?

The researchers applied their training approach to several existing VLMs, finding that the models' performance consistently improved after the training. That suggests that the navigation data from pathologists was useful across settings, Asadi said.

For Huang, that is the most important result. "The takeaway isn't our system," he said. "It's that the missing ingredient has been sitting in hospitals this whole time."

The study did not compare Pathology-o3 directly with human pathologists, but the researchers said that wasn't their aim.

"The right question isn't whether it beats a pathologist," Huang said. "It's whether a pathologist working with it catches more [cancer cases] and works faster." The current study did not address the latter question, either, but the team's next experiment is designed to test pathologists on the same cases with and without Pathology-o3, measuring what they catch and how long they take to do so.

The system's most plausible use is as a prescreening tool, Asadi said, but he stressed that the research has not yet shown that doctors who use it become faster or more accurate. Asadi wants an even tougher test: trials conducted across multiple hospitals that measure not just accuracy and speed but also pathologists' workloads. He wants the trails to assess the burden of false alarms from the AI algorithms and whether doctors recognize when the AI is wrong.

Importantly, cancer diagnoses can require information from multiple slides, stains and a patient's medical history, while the current system just reads one slide at a time. "I wouldn't claim it should diagnose on its own," Huang said.

This article is for informational purposes only and is not meant to offer medical advice.

'> AI trained to 'think' like human pathologists may be better at spotting cancer