While not as deep and complex as a Super Bowl half-time show, the bees’ moves, known as the “waggle” dance, convey very specific food foraging instructions to their nestmates. The direction the dancer moves explains to other bees which way to go, and the duration of the waggle dance, or the “run,” shows how far to go. Once other bees have been convinced to follow the directions, they are “recruited.” After receiving the instructions, these recruits leave the hive to find the food their sisters were so excited about.

Unfortunately, many of these recruited bees do not always successfully find the food they set out in search of. Margaret Couvillon, associate professor in the Department of Entomology in the College of Agriculture and Life Sciences, and her former Ph.D. student Laura McHenry wanted to find out why.

Trying to understand why waggle dances fail

Honeybees have had millions of years to perfect the waggle dance, so it may be surprising to learn that it doesn’t often work. Even though it was first described by scientists over 80 years ago, there is still a lot about the waggle dance that we don’t understand.

Couvillon has learned several interesting patterns related to this form of communication. One such observation was that bees have consistent, unique ways of dancing, meaning each bee has its own “style” that it adds to the communication. Could the success of the waggle dance be related to this uniqueness? Would bees that communicated similarly yield more successful recruits? Or is there some other factor at play? This study reveals the waggle to be a diverse form of communication that helps improve the likelihood that one bee can tell another where food can be found. The findings were recently published in Current Biology.

“Although the waggle dance itself is fascinating, my lab has additionally been intrigued about waggle dance miscommunication, or the hows and whys behind the failure of the dance recruitment,” Couvillon said.

To answer these questions, the Couvillon Lab devised an experiment utilizing clear-walled hives, video cameras, and a method of tagging bees so they could be tracked as individuals when they foraged and danced. Each hive included foragers who had been taught the location of an artificial food source. These trained foragers performed a waggle dance to teach others where this food was, effectively training a new set of recruits. If successful in locating the food, these recruits returned to teach other bees what they learned. Couvillon and her team hypothesized that bees with similar dance styles would more often successfully teach others how to find the food and communication that differed between bees would be less successful.

Whenever a new, tagged bee was observed at the food source, video of the hive was reviewed to determine which dancer had recruited that successful forager. This pattern of data collection allowed the researchers to track the dance the bees used, with each bee learning where the food was located from a slightly different telling. These successful dances were then compiled, and the run of each dance was measured and compared to the earlier dances. The pattern that emerged was not what the researchers expected.

The power of individuality

Based on the data from these dances, Couvillon and McHenry found that similar dance communication did not actually result in the most successful foraging, which was their original hypothesis. Dances that had a longer run, effectively telling the recruits to overshoot the food source, were more successful than dances describing similar, more accurate, distances. This pattern suggested that the “overshooting” instructions may have led to additional opportunities to find the food, once on the way past the food source and again on the way back to the hive. They theorized that the foragers having a second chance to find the food source increased the chance that they find it at all.

What does this mean for understanding the honeybee waggle dance? One takeaway is the importance of these unique communication styles, where individual dance mannerisms enhance communication success. If every bee communicated the same, the likelihood of foragers reaching the food would decrease as compared to having a diverse set of styles.

This study adds effective dance moves to the list of known benefits of individuality, showing that a diverse set of communication skills helps improve the likelihood that one bee can tell another where food can be found, all through dance.

“We’ve known for a while that behavioral and genetic diversity benefit honeybees, allowing for superior thermoregulation, disease resistance, growth, and foraging,” said Couvillon. “Now we have also seen that diverse communication enhances recruitment success.”

Called the Electronic Bee-Veterinarian, or EBV, the technology uses low-cost heat sensors and forecasting models to predict when hive temperatures may reach dangerous levels. The system provides remote beekeepers with early warnings, allowing them to take preventive action before their colonies collapse during extreme hot or cold weather or when the bees cannot regulate their hive temperature because of disease, pesticide exposure, food shortages, or other stressors.

“We convert the temperature to a factor that we are calling the health factor, which gives an estimate of how strong the bees are on a scale from zero to one,” said Shamima Hossain, a Ph.D. student in computer science at UCR and lead author of a paper explaining the technology.

This simplified metric — with a score of ‘one’ meaning the bees are at full strength — allows beekeepers unfamiliar with the underlying model to assess hive health quickly.

Boris Baer, a UCR professor of entomology, believes the technology could revolutionize beekeeping, which is essential to vast sectors of global agriculture. Honeybees pollinate more than 80 crops and contribute an estimated $29 billion annually to U.S. agriculture. Yet bee populations have declined due to various factors, including habitat loss, pesticide exposure, parasites, and climate change.

“Over the last year, the U.S. lost over 55% of its honeybee colonies,” Baer said, citing data from Project Apis m., which monitors beehive losses throughout the U.S. “We are experiencing a major collapse of bee populations, and that is extremely worrying because about one-third of what we eat depends on bees.”

Beekeepers now rely on their own judgment and manual inspections to detect problems, often leading to delayed interventions. With EBV, they can get real-time insights and predict conditions days in advance, significantly reducing labor costs, said Baer, who collaborated with Hossain and other scientists at UCR’s Bourns College of Engineering.

“People have dreamed of these sensors for a very long time,” Baer said. “What I like here is that this system is fully integrated into the hive setup that beekeepers already use.”

Temperature fluctuations are among the first responses to any kind of threats to a hive’s health. Honeybees maintain a precise internal hive temperature between 33 and 36 degrees Celsius (91.4-96.8°F), a requirement for proper brood development and colony survival, Baer said.

The EBV method is based on thermal diffusion equations and control theory, making its predictions interpretable to both scientists and beekeepers, Hossain said. The model uses temperature data collected from low-cost sensors installed inside the hive, feeding that information into an algorithm that predicts hive conditions several days in advance.

In tests conducted at UCR’s apiary, the EBV method analyzed data from 10 hives during initial development and later expanded to 25 hives. The technology has already proven its effectiveness, detecting conditions that required beekeeper intervention.

“When I looked at the dashboard and saw the health factor dropped below an empirical threshold, I contacted our apiary manager,” Hossain recalled. “When we went to check the hive, we found that there was actually something wrong, and they were able to take action to manage the situation.” Hyoseung Kim, an associate professor of electrical and computer engineering at UCR, explained that keeping costs low — under $50 per hive — is a high priority.

“There are commercial sensors available, but they are too expensive,” Kim said. “We decided to create a very cheap device using off-the-shelf components so that beekeepers can afford it.”

The research team is already working on the next phase, which is to develop automated hive climate controls that can be installed on hives and respond to EBV’s predictions, adjusting hive temperatures automatically.

“Right now, we can only issue warnings,” Hossain said. “But in the next phase, we are working on designing a system that can automatically heat or cool the hive when needed.”

The title of Hossain’s paper is “Principled Mining, Forecasting and Monitoring of Honeybee Time Series with EBV+” In addition to Hossain, Baer and Kim, the co-authors are Christos Faloutsos, professor of computer science at Carnegie Mellon University, and Vassilis Tsotras, professor of computer science and engineering at UCR.

All the authors are with UCR’s Center for Integrative Bee Research, one of the largest pollinator health research hubs in the nation.

Such is the case with a hybrid microscope, born at the Marine Biological Laboratory (MBL), that for the first time allows scientists to simultaneously image the full 3D orientation and position of an ensemble of molecules, such as labeled proteins inside cells. The research is published this week in Proceedings of the National Academy of Sciences.

The microscope combines polarized fluorescence technology, a valuable tool for measuring the orientation of molecules, with a dual-view light sheet microscope (diSPIM), which excels at imaging along the depth (axial) axis of a sample.

This scope can have powerful applications. For example, proteins change their 3D orientation, typically in response to their environment, which allows them to interact with other molecules to carry out their functions.

“Using this instrument, 3D protein orientation changes can be recorded,” said first author Talon Chandler of CZ Biohub San Francisco, a former University of Chicago graduate student who conducted this research partly at MBL. “There’s real biology that might be hidden to you from just a position change of a molecule alone,” he said.

Imaging the molecules in the spindle of a dividing cell — a longstanding challenge at MBL and elsewhere — is another example.

“With traditional microscopy, including polarized light, you can study the spindle quite nicely if it’s in the plane perpendicular to the viewing direction. As soon as the plane is tilted, the readout becomes ambiguous,” said co-author Rudolf Oldenbourg, a senior scientist at MBL. This new instrument allows one to “correct” for tilt and still capture the 3D orientation and position of the spindle molecules (microtubules).

The team hopes to make their system faster so that they can observe how the position and orientation of structures in live samples change over time. They also hope development of future fluorescent probes will enable researchers to use their system to image a greater variety of biological structures.

A Confluence of Vision

The concept for this microscope gelled in 2016 through brainstorming by innovators in microscopy who met up at the MBL.

Hari Shroff of HHMI Janelia, then at the National Institutes of Health (NIH) and an MBL Whitman Fellow, was working with his custom-designed diSPIM microscope at MBL, which he built in collaboration with Abhishek Kumar, now at MBL.

The diSPIM microscope has two imaging paths that meet at a right angle on the sample, allowing researchers to illuminate and image the sample from both perspectives. This dual view can compensate for the poor depth resolution of any single view, and illuminate with more control over polarization than other microscopes.

In conversation, Shroff and Oldenbourg realized the dual view microscope could also address a limitation of polarized light microscopy, which is that it’s difficult to efficiently illuminate the sample with polarized light along the direction of light propagation.

“If we had two orthogonal views, we could sense polarized fluorescence along that direction much better,” Shroff said. “We thought, why not use the diSPIM to take some polarized fluorescence measurements?”

Shroff had been collaborating at MBL with Patrick La Rivière, a professor at University of Chicago whose lab develops algorithms for computational imaging systems. And La Rivière had a new graduate student in his lab, Talon Chandler, whom he brought to MBL. The challenge of combining these two systems became Chandler’s doctoral thesis, and he spent the next year in Oldenbourg’s lab at MBL working on it.

The team, which early on included Shalin Mehta, then based at MBL, outfitted the diSPIM with liquid crystals, which allowed them to change the direction of input polarization.

“And then I spent a long time working through, what would a reconstruction look like for this? What is the most we can recover from this data that we are now starting to acquire?” Chandler said. Co-author Min Guo, then located at Shroff’s previous lab at NIH, also worked tirelessly on this aspect, until they had reached their goal of full 3D reconstructions of molecular orientation and position.

“There was tons of cross-talk between the MBL, the University of Chicago, and the NIH, as we worked this through,” Chandler said.

“The results show that glass fertilizers can be tailored to plant needs, slowly and sustainably releasing nutrients to boost productivity without harming soil quality,” says Danilo Manzani, a co-author of the study.

Over time, the use of agricultural chemicals has increased. In 2020, the Food and Agriculture Organization of the United Nations estimated that global demand for fertilizers would surpass 200 million metric tonnes. Fertilizers contain nitrogen, phosphorus and lower amounts of other elements like calcium. Unfortunately, the benefits of these nutrients are lost through leaching into groundwater and emissions into the air, necessitating frequent reapplication and creating downstream environmental problems like toxic algal blooms. A potential solution could come from tiny glass beads that previous researchers used to improve plant growth. To improve the efficiency of nutrient delivery, Manzani, Eduardo Ferreira and colleagues developed a water-soluble, multicomponent glass fertilizer designed for controlled nutrient release.

The researchers synthesized glass consisting of several micro- and macronutrients, such as phosphorus, potassium and calcium. They ground the glass into small (less than 0.85 millimeters wide) and large (0.85 to 2 millimeters wide) particles. In an initial test, the particles were added to either water or a buffer solution that mimicked soil conditions. They found that each nutrient released from both sizes of glass particles and diffused into the solutions steadily over 100 hours with minor fluctuations.

They then applied a nutrient solution or different amounts of the glass beads to soil seeded with a typical lawn and fairway grass, and they compared the plants’ growth in the two treatments. The nutrient solution, which was applied only once, immediately stimulated plant growth, but the effect quickly diminished. However, the single application of glass fertilizer sustained plant growth regardless of particle size, though overall growth depended on the bead dose.

Manzani, Ferreira and colleagues also examined the possible ecotoxicity of the glass fertilizer by exposing lettuce and onion seeds to the beads. Seeds exposed to glass fertilizer had roughly the same germination rate and cell health as those never exposed or those treated with soluble nutrients. The researchers say that these results indicate an efficient and sustained alternative to conventional fertilizers with lower environmental impact.

The authors acknowledge funding from the São Paulo Research Foundation; Center for Research, Technology, and Education in Vitreous Materials; National Council for Scientific and Technological Development (Conselho Nacional de Desenvolvimento Científico e Tecnológico); Coordination for the Improvement of Higher Education Personnel; and Funding Authority for Studies and Projects.

Hair conditioners typically contain 20-30 ingredients, many derived from petroleum and oleochemicals, raising concerns about sustainability and environmental impact. A new study published in Science Advances, demonstrates that micellar lignin gels can effectively stabilize emulsions with natural oils, reducing the need for synthetic surfactants and complex stabilizers commonly used in commercial formulations. The research team, led by Mika Sipponen at Stockholm University, sought to explore lignin, a common and renewable component in wood biomass, as a multifunctional component for hair conditioning.

“Our findings highlight lignin’s potential as a stabilizer in oil-in-water emulsions, enabling a more natural and sustainable approach to hair conditioning,” says Mika Sipponen. “By using wood-derived lignin directly without any chemical modification, we not only simplify the ingredient list but also eliminate the need for organic solvents, making the process more eco-friendly.”

Comparable to commercial hair conditioners

The lignin gel-based conditioner was tested against a commercial hair conditioner, showing comparable emulsion stability, viscosity, and conditioning performance. A formulation with 6 percent coconut oil effectively lubricated damaged hair, reducing wet combing force by 13 percent, as confirmed by combing force measurements and multiscale microscopy analysis. Importantly, the product was easily rinsed off from paper and skin with cold water despite its dark color, demonstrating practical usability.

New opportunities in cosmetics and food

Ievgen Pylypchuk, who has been instrumental in developing lignin gel as a versatile platform material, highlights its broader potential: “Our lignin gel technology extends beyond personal care applications. Its unique ability to stabilize emulsions and interact with various biomolecules opens opportunities in cosmetics, food, and even biomedical formulations, offering a sustainable alternative to conventional ingredients.”

This innovation paves the way for greener haircare solutions that align with growing consumer demand for sustainable personal care products. The researchers anticipate further exploration of lignin-based formulations for broader applications in the personal care industry.

“We’ve figured out a way for robots to behave more like a material,” said Matthew Devlin, a former doctoral researcher in the lab of UCSB mechanical engineering professor Elliot Hawkes, and the lead author of a paper published in the journal Science. Composed of individual, disk-shaped autonomous robots that look like small hockey pucks, the members of the collective are programmed to assemble themselves together into various forms with different material properties.

Of particular interest to the research team was the challenge of creating a robotic material that could both be stiff and strong, yet be able to flow when a new form is needed. Rather than responding to exterior forces to attain a form, robotic materials ideally would respond to internal signals, Hawkes explained, able to take a shape and hold it, “but also able to selectively flow themselves into a new shape.”

For inspiration, the researchers tapped previous work by Otger Campàs, a former UCSB professor and currently the director of the Physics of Life Excellence Cluster at TU Dresden, on how embryos are physically shaped. “Living embryonic tissues are the ultimate smart materials,” he said. “They have the ability to self-shape, self-heal and even control their material strength in space and time.” While at UCSB, his laboratory discovered that embryos can melt like glass to shape themselves. “To sculpt themselves, cells in embryos can make the tissues switch between fluid and solid states; a phenomenon known as rigidity transitions in physics,” he added.

During the development of an embryo, cells have the remarkable ability to arrange themselves around each other, turning the organism from a blob of undifferentiated cells into a collection of discrete forms — like hands and feet — and of various consistencies, like bones and brain. The researchers concentrated on enabling three biological processes behind these rigidity transitions: the active forces developing cells apply to one another that allow them to move around each other; the biochemical signaling that allow these cells to coordinate their movements in space and time; and their ability to adhere to each other, which ultimately lends the stiffness of the organism’s final form.

In the world of robots, the intracellular forces translate to inter-unit tangential force, enabled by eight motorized gears along each robot’s circular exterior, which allow them to move around each other, pushing off each other, even in tightly packed spaces.

The biochemical signaling, meanwhile, is akin to a global coordinate system. “Each cell ‘knows’ its head and tail, so then it knows which way to squeeze and apply forces,” Hawkes explained. In this way, the collective of cells manages to change the shape of the tissue, such as when they line up next to each other and elongate the body.

In the robots, this feat is accomplished by light sensors on the top of each robot, with polarized filters. When light is shone on these sensors, the polarization of the light tells them which direction to spin its gears and thus how to change shape. “You can just tell them all at once under a constant light field which direction you want them to go, and they can all line up and do whatever they need to do,” Devlin added.

For the cell-cell adhesion the researchers used magnets incorporated into the perimeter of the robotic units, magnets that could be turned to attract any other robot.

In putting the robots through their paces, the researchers found that signal fluctuations — variations in the signals sent to the robots — played a critical role in their ability to take the necessary shapes and formations. “We had previously shown that in living embryos, the fluctuations in the forces that cells generate are key to turning a solid-like tissue into a fluid one. So, we encoded force fluctuations in the robots,” said Campàs.

In the robot collective, the interaction between signal fluctuations and inter-unit forces is the difference between a tightly packed, unmoving collective and a more fluid one. “Basically, as you increase both of those, especially fluctuations, you get a more flowing material,” Devlin said. This allows the collective to change shape. Once in formation, switching off the force fluctuations rigidifies the collective again.

Importantly, these signal fluctuations make it possible for the robot collective to achieve their shape and strength changes with less average power than if the signal were constantly on and the robots were all pushing on each other continuously. “It’s an interesting result that we did not set out looking for, but discovered once we started gathering data on the robot behaviors,” Hawkes said. This is important, he added, for designing robots that may have to run on limited power budgets.

With all this in mind, the researchers were able to tune and control the group of robots to act like a smart material: sections of the group would turn on dynamic forces between robots and fluidize the collective, while in other sections the robots would simply hold to each other to create a rigid material. Modulating these behaviors across the group of robots and over time allowed the researchers to create robotic materials that support heavy loads but can also reshape, manipulate objects, and even self-heal.

Currently, the proof-of-concept robot collective comprises a small number (20) of relatively large units, but simulations conducted by former Campàs laboratory postdoctoral fellow Sangwoo Kim, who is now an assistant professor at EPFL, indicate the system can be scaled to larger numbers of miniaturized units, for a more materials-like aspect.

Beyond robotics, according to the paper, this and robot collectives like it could “enable the study of phase transitions in active matter, the properties of active mechanics in particulate systems and potentially help define hypotheses for biological research.” Combined with current controls and machine learning strategies, working with these robot collectives could yield emergent capabilities in robotic materials that have yet to be discovered and understood.

Their study, published in Endangered Species Research, demonstrates how microphones combined with ML and traditional aerial survey methods can help track right whale populations in Cape Cod Bay, a crucial feeding ground where the whales gather each spring.

To track this endangered species, researchers rely on costly and dangerous surveys by airplanes, or use sound recordings to identify their presence, or absence.

“Using sound recordings to monitor whale populations isn’t new,” said lead author Marissa Garcia of the Cornell Lab of Ornithology’s K. Lisa Yang Center for Conservation Bioacoustics. “What makes our study unique is that we were able to take those recordings and go beyond getting information on the presence or absence of whales to getting an approximate number of whales in an area.”

The team set out an array of marine autonomous recording units (MARU) across Cape Cod Bay to capture right whale sounds.

Following deployment of the MARUs, the team trained, validated and applied a deep-learning model that could automatically detect right whale sounds with 86% precision.

“By analyzing their distinctive upcall vocalizations, we can detect their presence continuously, day and night,” Garcia said. “This kind of round-the-clock monitoring that results from passive acoustic monitoring just isn’t possible with traditional aerial surveys, which can only happen in daylight hours and in good weather.”

Garcia says there’s still some uncertainty in the counts that the team needs to address in future research, but the team is optimistic that monitoring whale vocalizations holds promise for estimating the abundance of right whales to aid in conservation and management efforts.

Having the ability to expand monitoring efforts across larger areas of the ocean will help scientists better assess the species’ population numbers across the full extent of its range. Garcia said right whales have been traditionally thought of as a conservation challenge in New England, but right whales are found all along the East Coast.

“Using passive acoustic data and deep-learning tools, we can expand the area we can safely monitor and keep track of this critically endangered species,” Garcia said.

The work comes at a critical time for North Atlantic right whales, whose population has declined to fewer than 370 individuals due to ship strikes, fishing gear entanglement and changing ocean conditions affecting their food sources.

Analogical reasoning is the ability to draw a comparison between two different things based on their similarities in certain aspects. It is one of the most common methods by which human beings try to understand the world and make decisions. An example of analogical reasoning: cup is to coffee as soup is to (the answer being: bowl)

Large language models like GPT-4 perform well on various tests, including those requiring analogical reasoning. But can AI models truly engage in general, robust reasoning or do they over-rely on patterns from their training data? This study by language and AI experts Martha Lewis (Institute for Logic, Language and Computation at the University of Amsterdam) and Melanie Mitchell (Santa Fe Institute) examined whether GPT models are as flexible and robust as humans in making analogies. ‘This is crucial, as AI is increasingly used for decision-making and problem-solving in the real world’, explains Lewis.

Comparing AI models to human performance

Lewis and Mitchell compared the performance of humans and GPT models on three different types of analogy problems:

1. Letter sequences — Identifying patterns in letter sequences and completing them correctly.
2. Digit matrices — Analyzing number patterns and determining the missing numbers.
3. Story analogies — Understanding which of two stories best corresponds to a given example story.

A system that truly understands analogies should maintain high performance even on variations

In addition to testing whether GPT models could solve the original problems, the study examined how well they performed when the problems were subtly modified. ‘A system that truly understands analogies should maintain high performance even on these variations’, state the authors in their article.

GPT models struggle with robustness

Humans maintained high performance on most modified versions of the problems, but GPT models, while performing well on standard analogy problems, struggled with variations. ‘This suggests that AI models often reason less flexibly than humans and their reasoning is less about true abstract understanding and more about pattern matching’, explains Lewis.

In digit matrices, GPT models showed a significant drop in performance when the position of the missing number changed. Humans had no difficulty with this. In story analogies, GPT-4 tended to select the first given answer as correct more often, whereas humans were not influenced by answer order. Additionally, GPT-4 struggled more than humans when key elements of a story were reworded, suggesting a reliance on surface-level similarities rather than deeper causal reasoning.

On simpler analogy tasks, GPT models showed a decline in performance decline when tested on modified versions, while humans remained consistent. However, for more complex analogical reasoning tasks, both humans and AI struggled.

Weaker than human cognition

This research challenges the widespread assumption that AI models like GPT-4 can reason in the same way humans do. ‘While AI models demonstrate impressive capabilities, this does not mean they truly understand what they are doing’, conclude Lewis and Mitchell. ‘Their ability to generalize across variations is still significantly weaker than human cognition. GPT models often rely on superficial patterns rather than deep comprehension.’

This is a critical warning for the use of AI in important decision-making areas such as education, law, and healthcare. AI can be a powerful tool, but it is not yet a replacement for human thinking and reasoning.