The technology could give small satellites far greater flexibility in space. Instead of relying on separate fuel systems for different types of maneuvers, future spacecraft could use a single propellant to perform both rapid movements and slow, highly controlled adjustments.

At the center of the approach is a specialized fuel that works with both chemical and electric propulsion systems. Until now, these technologies have typically required separate propellants and hardware, adding weight and complexity.

“If you can have chemical and electrical propulsion in one small package, it’s the best of both worlds,” says Amelia Bruno, a former postdoc in MIT’s Department of Aeronautics and Astronautics (AeroAstro). “This opens the door for small satellites to do even more science, more observations, and more interesting missions, all on a smaller and cheaper platform.”

Bruno is the lead author of a new study published in the Journal of Propulsion and Power. The research demonstrates that a “green monopropellant” originally developed by the U.S. Air Force for chemical propulsion can also successfully power miniature electric thrusters known as electrospray thrusters.

Combining Chemical and Electric Space Propulsion

Electrospray thrusters are tiny rocket engines, roughly the size of a dime. They use electric fields to charge particles in a liquid propellant and then eject those particles into space, creating thrust.

These thrusters are extremely fuel-efficient and are well suited for gradual, precise maneuvers. For example, they can slowly push a spacecraft through long interplanetary journeys while consuming very little fuel.

Chemical thrusters serve a different purpose. They deliver powerful bursts of thrust that allow spacecraft to quickly accelerate, decelerate, climb, descend, or change position.

By identifying a propellant capable of powering both systems, MIT researchers believe they can significantly expand the capabilities of small satellites.

The team is currently working with NASA on the Green Propulsion Dual Mode mission, a briefcase-sized CubeSat equipped with one chemical thruster and four electrospray thrusters. All of them will draw fuel from a single tank. The mission will be the first attempt to test this type of dual-mode propulsion system on a small spacecraft.

If successful, the technology could help small satellites venture far beyond Earth orbit.

“We could send CubeSats to Mars, or the asteroid belt, where they could make the journey slowly, using electrospray thrusters,” says study co-author Paulo Lozano, the Miguel Alemán Velasco Professor of Aeronautics and Astronautics at MIT. “You could then use your chemical thrusters to quickly move to look at interesting features. You could have a lot more flexibility to do a lot more things.”

Why Ionic Liquid Propellants Matter

Lozano’s laboratory develops, manufactures, and tests electrospray propulsion systems for satellites ranging in size from a lunchbox to a small carry-on suitcase.

Compared with larger spacecraft, these compact satellites are much less expensive to launch. Their smaller size, however, requires equally compact propulsion systems.

Electrospray thrusters fit that requirement well. The devices created in Lozano’s lab are about the size of a thumbnail. Each thruster sits above a reservoir containing an ionic liquid propellant. When connected to a battery, an electric charge is applied to ions within the liquid. Those charged particles are then expelled through tiny openings in the thruster, producing thrust.

Over the past decade, Lozano’s group has tested numerous designs under different operating conditions and with a variety of ionic liquid fuels.

“Ionic liquids are very stable and can even remain a liquid in space, which not a lot of materials can do,” Bruno says. “And it’s basically a sea of ions, which is why we base our technology around it, so we can pull those ions out into an electrospray.”

MIT researchers have also collaborated with the U.S. Air Force, which developed a new ionic liquid fuel known as the Advanced SpaceCraft Energetic Non-Toxic propellant (ASCENT). The propellant was originally designed for chemical propulsion systems.

ASCENT was created as a safer alternative to hydrazine, the highly toxic fuel traditionally used in many spacecraft propulsion systems.

“ASCENT happens to be an ionic liquid mixture,” Bruno says. “And we said, hey, that’s the stuff we typically use. Theoretically, this should work. Let’s go figure out how.”

Testing ASCENT in Electrospray Thrusters

To evaluate the fuel, Bruno, Lozano, and former MIT graduate student Matthew Corrado conducted a series of experiments using electrospray thrusters powered by ASCENT.

Each thruster was attached to a small cube-shaped reservoir approximately the size of a LEGO brick. Researchers filled each reservoir with one gram of ASCENT, a liquid with a viscosity similar to baby oil.

The thrusters were mounted on opposite sides of a CubeSat positioned on a custom magnetic levitation test platform known as the MagLev. The setup is located inside a large vacuum chamber that can recreate conditions similar to those found in space.

During testing, the researchers remotely varied the voltage supplied to the thrusters. The resulting electrospray generated enough force to spin the CubeSat like a floating top.

By measuring the generated thrust and operating the thrusters continuously for periods of up to 100 hours, the team was able to assess the fuel’s performance and efficiency.

The results showed that ASCENT successfully powered the electrospray thrusters. The fuel performed on par with conventional ionic liquid propellants typically used in electric propulsion systems.

“Compared to our normal electrospray propellants, ASCENT can provide similar performance in terms of thrust,” Bruno says. “Now that we know our thrusters work with ASCENT, we can start thinking of all the ways we can make them even better.”

NASA Mission Will Test Shared Fuel Tank in Space

With ASCENT now proven capable of supporting both chemical and electric propulsion, researchers envision future spacecraft carrying a single fuel tank to power both systems.

That concept will soon face its first real-world test through NASA’s Green Propulsion Dual Mode mission, which is scheduled for launch in November.

“This will be the first time that a satellite will have a shared propellant tank,” says Lozano.

Beyond deep-space exploration, the technology could also improve missions closer to Earth. Lozano points to weather and climate monitoring as one potential application.

“Say there’s a storm coming, and you’d want to deploy your constellation of small satellites to observe over one location,” he says. “You could choose to send them quickly or slowly depending on the nature of the observation. And the only way to do that is if you have two propulsion systems, which is now possible.”

This research was supported in part by NASA.

Researchers led by Suketu Patel put several leading AI models through a well-known psychology experiment called the Stroop task. The results revealed a significant difference between how AI systems process information and how the human brain manages attention.

What Is the Stroop Task?

The Stroop task is a classic psychological test that has been used for decades to study attention, concentration, and self-control.

In the test, color words such as “red,” “blue,” or “green” are displayed in colored ink. Sometimes the word and the ink color match. For example, the word “red” might appear in red ink. Other times they conflict, such as the word “red” printed in blue ink.

Participants are asked to name the color of the ink rather than read the word itself.

That sounds simple, but it creates a challenge because reading words is an automatic habit for most people. The brain must suppress the urge to read the word and instead focus on identifying the ink color.

Psychologists often use the task to measure what is known as executive control, a set of mental processes that helps people regulate attention, resist distractions, and stay focused on goals.

Testing AI Attention

The researchers wanted to see whether modern large language models (LLMs) handle this challenge in the same way humans do.

LLMs are the AI systems behind tools such as ChatGPT, Claude, and Gemini. They are trained on enormous amounts of text and learn patterns in language, allowing them to generate responses that often appear remarkably human.

When given short lists containing five color words, the AI systems generally performed well, even when the words and colors did not match.

However, the picture changed dramatically as the lists became longer.

GPT-4o achieved 91% accuracy when working with five words. At ten words, its accuracy fell to 57%. When the list expanded to forty words, accuracy dropped to just 15%.

Claude 3.5 Sonnet maintained stable performance through lists of twenty words but then experienced a sharp decline, falling to 24% accuracy with forty-word lists.

The researchers observed similar patterns in GPT-5, Claude Opus 4.1, and Gemini 2.5.

When AI Loses Focus

The challenge became even more difficult when matching and mismatched color words appeared together in the same list.

Under those conditions, performance deteriorated further. Accuracy for the mismatched items dropped to nearly zero in some cases.

According to the researchers, the AI models had trouble maintaining the instruction to identify ink colors. Instead, they increasingly defaulted to reading the words themselves.

In other words, the systems appeared unable to consistently suppress the response they had been most heavily trained to produce.

This finding is particularly interesting because humans face a similar conflict. People are generally much better at reading words than naming ink colors. Yet despite this bias, most individuals can maintain high accuracy and stable performance even when confronted with long lists of conflicting words and colors.

Human Attention vs. Machine Attention

The study highlights an important distinction between human and artificial intelligence.

Although modern AI systems can produce impressive language and reasoning capabilities, their underlying mechanisms differ from the attention processes found in biological brains.

Humans can often sustain focus on a specific goal while filtering out competing information. The results suggest that current AI models may struggle with this type of cognitive control when tasks become increasingly demanding.

The researchers argue that the performance collapse seen in these experiments points to fundamental limitations in today’s large language models. While AI can sometimes mimic human behavior, its ability to maintain attention appears to operate very differently from the way people do.

The findings offer a reminder that even the most advanced AI systems still have weaknesses, particularly when tasks require them to resist distractions and stay focused over extended sequences of information.

The achievement gives scientists a new way to examine how the brain and body work together to produce complex actions, including walking and flying. It also opens the door to broader studies of the core rules that govern nervous systems.

“We can see all of the neurons and their connections as a complete unit for the first time and ask, ‘What do we learn from that?'” said study co-senior author Rachel Wilson, the Joseph B. Martin Professor of Basic Research in the Field of Neurobiology in the Blavatnik Institute at HMS.

First Complete Fruit Fly Brain and Body Wiring Map

The new map of neural connections, known as a connectome, extends a previously published fruit fly brain connectome by adding the fly’s spinal cord equivalent, called the nerve cord.

“It is really important to have a central nervous system connectome that is as complete as possible so we can link up the brain and body and start thinking about behavior holistically,” said study co-senior author Wei-Chung Allen Lee, associate professor of neurobiology at HMS and HMS professor of neurology at Boston Children’s Hospital.

When the team studied the connectome, they discovered that many fruit fly behaviors appear to be directed by local neural circuits in the relevant body parts, rather than by one central command area in the brain.

The full connectome is now freely available online, giving researchers around the world a powerful new resource for neuroscience studies. The work, published June 8 in Nature, received support in part from U.S. federal funding, including the BRAIN Initiative (Brain Research Through Advancing Innovative Neurotechnologies), National Institutes of Health, and National Science Foundation.

Why Fruit Flies Matter in Neuroscience

One of neuroscience’s major unanswered questions is how neurons in the brain and body connect and coordinate to generate behavior. The fruit fly Drosophila melanogaster is a valuable model for exploring that problem.

Fruit flies are simple to breed and keep in the lab. Although their nervous system contains only about 160,000 neurons, they can still perform complex behaviors such as navigating, interacting socially, learning, and reacting to sensory signals. They also have what Lee describes as an incredibly sophisticated genetic toolkit, which allows scientists to access, control, and record activity from single neurons or groups of neurons.

In 2024, the FlyWire Consortium, led by Mala Murthy and Sebastian Seung at Princeton, who are also co-authors of the new study, published a complete connectome of a fruit fly brain. At the same time, Lee and his colleagues were building a connectome of the fruit fly nerve cord, which controls the legs, wings, and other appendages while also processing sensory information.

“The brain and nerve cord connectomes are each useful on their own, but until you can bridge the two, it’s hard to understand how information moves between the brain and the body,” said co-first author Helen Yang, a research fellow in neurobiology in the Wilson Lab.

Co-first author Alexander Bates, also a research fellow in neurobiology in the Wilson Lab, noted that the brain holds most of the neurons, but the nerve cord contains neurons that are “some of the most useful” because they are tied to sensation, movement, and functions that are often easier to interpret.

Connecting the Brain to the Nerve Cord

The FlyWire team was eager to shift toward the brain and neural cord, or BANC, dataset imaged in the Lee Lab, said co-senior author Murthy, the Karol and Marnie Marcin ’96 Professor of Neuroscience at Princeton and director of the Princeton Neuroscience Institute (PNI).

“The new connectome represents a major advance for the field, with the ability to understand how circuits in the brain receive feedback from and control the actions of the body,” she said.

“For the first time, we can follow information flow from sensation to action across an entire nervous system,” added co-author Arie Matsliah of the PNI.

How Scientists Built the Connectome

To create the connectome, researchers sliced a single fruit fly into thousands of extremely thin serial sections. They then used electron microscopy to capture millions of images showing neurons and their connections. AI tools helped align those images and assemble them into a unified 3D map.

The finished connectome shows how each neuron connects with other neurons in the brain and nerve cord at the level of individual synapses. The map does not cover the fly’s entire body, but the researchers used identifiable neurons and previous scientific literature to link central nervous system neurons with neurons in many appendages and sensory organs, effectively “embodying” the connectome.

Lee said scientists can use the connectome to develop new hypotheses for lab experiments. He compares it to having detailed Google Maps information while planning a route.

“The connectome has shown us that most of our hypotheses are too simple. Now, we can develop more complex hypotheses and move forward with experiments to test them,” Lee said.

A Surprise About How Movement Is Controlled

The researchers have already used the connectome to study motor control, especially how a fly moves its legs and other body parts.

A long-standing idea in neuroscience holds that the brain acts as a centralized controller that decides which actions an animal will perform. The fruit fly connectome pointed to a different answer.

The team found that motor control in fruit flies mostly occurs locally. For instance, movement of one leg is mainly governed by the neural circuits for that leg. Those circuits then communicate with circuits for the other legs to produce coordinated actions such as walking.

The same pattern appeared in circuits linked to the fly’s wings, mouth, and other body parts. The researchers also found that motor circuits connect with other circuit types, including those in the visual and endocrine systems, which supply extra information that helps shape behavior.

“Our findings suggest that control for actions is highly distributed in local modules that link up and work together in different ways,” Bates said.

What Comes Next for Connectome Research

The researchers say the connectome could support many future lines of investigation. Yang compares it to the Human Genome Project, another large-scale open resource that has been used in many different ways.

Soon, the team plans to add more information to the connectome, including details about neuropeptides, the small, protein-like molecules that neurons use to communicate.

The connectome may also reveal basic principles that apply to nervous systems across species, including humans. Bates said many discoveries from fruit fly neuroscience have carried over from invertebrates to mammals, including findings related to navigation, olfaction, and memory.

Another goal is “to bring full-connectome mapping to much more complex organisms,” said Matsliah. He noted that progress in AI, computing, and open collaborative science is making this kind of research increasingly possible.

A major question now is whether the distributed neural control observed in fruit flies is also found in other animals. Lee is currently investigating that possibility in mice.

“I would be shocked if this is unique to the fly,” Yang said. “We don’t have this level of resolution in other animals, but we know that they have a lot of these local circuits.”

Lessons for Artificial Intelligence

The work could also have implications for artificial intelligence. The connectome offers real biological data that may help guide the design of artificial agents that move through virtual worlds, systems that are increasingly used to study intelligence and improve AI training.

“One thing that always amazes me is that this tiny little fly does a hell of a lot; even our best AI agents and robots can’t do everything that a fly does,” Yang said. “There may be lessons for AI in how the nervous system is organized.”

Authorship, funding, disclosures

Jasper S. Phelps and Minsu Kim are also co-first authors of the study. Jan Drugowitsch is co-senior author. Additional authors include Zaki Ajabi, Eric Perlman, Kevin M. Delgado, Mohammed Abdal Monium Osman, Christopher K. Salmon, Jay Gager, Benjamin Silverman, Sophia Renauld, Farzaan Salman, Janki Patel, Matthew F. Collie, Jingxuan Fan, Diego A. Pacheco, Yunzhi Zhao, Wenyi Zhang, Laia Serratosa Capdevila, Ruairí J.V. Roberts, Eva J. Munnelly, Nina Griggs, Helen Langley, Borja Moya-Llamas, Zuoyu Zhang, Ryan T. Maloney, Szi-chieh Yu, Amy R. Sterling, Marissa Sorek, Krzysztof Kruk, Nikitas Serafetinidis, Serene Dhawan, Finja Klemm, Paul Brooks, Ellen Lesser, Jessica M. Jones, Sara E. Pierce-Lundgren, Su-Yee Lee, Yichen Luo, Andrew P. Cook, Theresa H. McKim, Dimitrios Stasi Giakoumas, Benjamin Gorko, Emily C. Kophs, Tjalda Falt, Alexa M. Negron-Morales, Austin Burke, James Hebditch, Kyle P. Willie, Ryan Willie, Sergiy Popovych, Nico Kemnitz, Dodam Ih, Kisuk Lee, Ran Lu, Akhilesh Halageri, J. Alexander Bae, Ben Jourdan, Gregory Schwartzman, Damian D. Demarest, Emily Behnke, Doug Bland, Anne Kristiansen, Jaime Skelton, Tom Stocks, Dustin Garner, Anthony Hernandez, Sandeep Kumar, The BANC-FlyWire Consortium, Kevin C. Daly, Sven Dorkenwald, Forrest Collman, Marie P. Suver, Lisa M. Fenk, Michael J. Pankratz, Zepeng Yao, Stephen J. Huston, Tomke Stürner, Gregory S.X.E. Jefferis, Katharina Eichler, Andrew M. Seeds, Stefanie Hampel, Sweta Agrawal, Tatsuo S. Okubo, Meet Zandawala, Thomas Macrina, Diane-Yayra Adjavon, Jan Funke, John C. Tuthill, Anthony Azevedo, and Benjamin L. de Bivort.

Funding was provided by the National Institutes of Health (grants R01NS121874; RF1MH117808; U19NS118246; U24NS126935; RF1MH117815; K99NS129759; R00NS117657; R01NS102333; RF1NS128785; R01NS140174; UM1NS132253; U24NS13992; RF1MH128840; R01NS121911; T32GM144273; R01DK139131; R25NS080687), a Sir Henry Wellcome Postdoctoral Fellowship (222782/Z/21/Z), a Smith Family Foundation Odyssey Award, a Harvard/MIT Joint Research Grant, an HHMI Life Sciences Research Foundation Postdoctoral Fellowship (PJ100000343), a New York Stem Cell Foundation Robertson Neuroscience Investigator Award, the Deutsche Forschungsgemeinschaft (ZA1296/1-1; EXC2151-390873048; PA787/7-3; PA787/9-3), the Nevada IDeA Network of Biomedical Research Excellence (GM103440), the National Science Foundation (2127379; 2014862), the Japan Society for the Promotion of Science (KAKENHI 25K00370), the Japan Science and Technology Agency (ASPIRE JPMJAP2302; CRONOS JPMJCS24K2), an HHMI Gilliam Fellowship (GT15790), the Max Planck Society, the Shanahan Family Foundation, a Kempner Graduate Fellowship, the Medical Research Council (MC_EX_MR/T046279/1), the Alice and Joseph Brooks Fund, and the Beijing Natural Science Foundation (IS23084). The authors also acknowledge that the work benefited from the O2 High-Performance Compute Cluster, supported by the Research Computing Group at HMS.

Harvard University filed a patent application for GridTape (WO2017184621A1) on behalf of the inventors, including W. Lee, and negotiated licensing agreements with interested partners. Macrina, Popovych, Kemnitz, Ih, K. Lee, Lu, Halageri, Bae, and Seung declare financial interest in Zetta AI. Seung declares financial interest in Memazing, Inc. Capdevila, Roberts, Langley, Munnelly, Griggs, and Moya-Llamas declare financial interest in Aelysia Ltd. Perlman is a principal of Yikes LLC.