By constructing computational models, or digital versions of the bee's brain, researchers have discovered how the way bees move during flight can help shape visual inputs and generate unique electrical messages within the brain.
These movements generate neural signals that allow honeybees to easily and efficiently identify predictable features of the surrounding world. This ability means that bees exhibit significant accuracy in learning and recognizing complex visual patterns in flight, such as those found in flowers.
This model not only helps to improve an understanding of how bees learn and recognize complex patterns through movement, but also paves the way for next-generation AI. Rather than relying on large-scale computing power, it shows that future robots could become smarter and more efficient by using movement to gather information.
Professor James Marshall, director of the Center for Machine Intelligence at the University of Sheffield and senior author of the study, said: “This study demonstrated that even the smallest people in the brain can use movement to perceive and understand the world around them.
“Utilizing the best designs of nature for intelligence opens the doors for next-generation AI and drives advances in robotics, self-driving cars and real-world learning.”
This study, a collaboration with Queen Mary University in London, is published in the journal today Elif. This is based on the team's previous research into how bees use active vision. This is a process in which movement helps to collect and process visual information. Their previous research has observed how bees fly around and examine specific patterns, but this new study provides a deeper understanding of the underlying brain mechanisms driving their behavior.
Bees' sophisticated visual pattern learning abilities, such as distinguishing human faces, have long been understood. However, the findings shed new light on how pollinators navigate the world with such seemingly simple efficiency.
Dr. Hadi Mabdi, lead author and researcher at the University of Sheffield, said: “Our previous research has found that bees solve visual puzzles using smart scan shortcuts.
“The bee brain model shows that its neural circuits are optimized to process visual information through positive interactions with flight movements in the natural environment, rather than alone, supporting theories that arise from how intelligence works from how the brain, body and environment work together.
“We learned that even though bees have a brain that is not larger than sesame seeds, they are not only seen in the world, but are actively shaped through movement. It's a beautiful example of how actions and perceptions are intertwined to solve complex problems with minimal resources. It has a huge impact on both biology and AI.”
This model shows that bee neurons are finely tuned to specific directions and movements as brain networks gradually adapt to repeated exposure to various stimuli and improve responses without relying on association or reinforcement. This allows the bee's brain to adapt to the environment simply by observing while flying, without needing immediate rewards. This means that the brain is very efficient and uses only a few active neurons to recognize things and preserve both energy and processing forces.
To validate their computational models, researchers were exposed to the same visual challenges that real honeybees encounter. In a pivotal experiment, the model was tasked with distinguishing between the “plus” sign and the “multiplied” symbol. This model significantly improved performance when mimicking the true bee strategy of scanning only the bottom half of the pattern. This is the behavior observed by the research team in previous research.
Even in a small network of artificial neurons, the model successfully demonstrated how bees recognize human faces, highlighting the strength and flexibility of visual processing.
Professor Lars Citta, Professor of Sensation and Behavioral Ecology at Queen Mary University, London, added: “Scientists have been fascinated by the question of whether brain size predicts the intelligence of animals. However, such speculations are meaningless unless you know the neural calculations that support a particular task.
“Here we determine the minimum number of neurons required for difficult visual identification tasks, and discover that numbers are surprisingly small even for complex tasks such as human face recognition. Therefore, insect microbrain is highly computational.”
Professor Mikko Zusola, professor of systems neuroscience at the Institute of Biological and Neuroscience at the University of Sheffield, said:
“Our new model extends this principle to bee's higher-order visual processing, revealing how behavior-driven scans are compressed and learnable neural codes are created. These findings support a unified framework where perception, action, and brain dynamics co-evolve to solve complex visual tasks with minimal resources – providing powerful insights for both biology and AI.”
By summarizing the findings from how insects behave, how their brains function, and what computational models show, this study shows how studying the brains of small insects can uncover the fundamental rules of intelligence. These findings not only provide a deeper understanding of cognitively, but also have great significance in the development of new technologies.
reference: Maboudi H, Roper M, Guiraud MG, and other neuromorphologic models of active vision demonstrate how spatiotemporal encoding in Lobula neurons can help bee pattern recognition. Elief. 2025. doi: 10.7554/Elife.89929
This article has been republished from material. Note: Materials may have been edited for length and content. For more information, please contact the source quoted. You can access the press release publishing policy here.
