Mosquitoes aren’t chasing each other—they’re all drawn to the same invisible signals that lead straight to you.
After observing hundreds of mosquitoes circling a human subject and analyzing roughly 20 million data points, researchers from Georgia Tech and the Massachusetts Institute of Technology developed a mathematical model that predicts how female mosquitoes locate and approach people to feed.
This work provides the first detailed visualization of mosquito flight behavior and offers measurable insights that could improve trapping and control methods. Mosquitoes are more than just a nuisance. They spread diseases such as malaria, yellow fever, and Zika, which together lead to more than 700,000 deaths each year.
The team also created an interactive public website that allows users to explore mosquito movement and behavior.
Tracking Mosquito Movement With 3D Cameras
To study how mosquitoes navigate, the researchers used 3D infrared cameras to monitor how the insects responded to visual signals and carbon dioxide around objects. They then placed a person inside a controlled chamber, changed his clothing colors, and recorded how mosquitoes moved around him.
The findings, published in Science Advances, focused on female Aedes aegypti mosquitoes (also called yellow fever mosquitoes), a species found across the southeastern United States, California, and many regions worldwide.
Mosquito Swarms Driven by Shared Signals
The data suggests that mosquitoes do not gather by following one another. Instead, each insect independently reacts to the same environmental cues, which leads them to arrive in the same place at roughly the same time.
“It’s like a crowded bar,” said David Hu, a professor in Georgia Tech’s George W. Woodruff School of Mechanical Engineering and the School of Biological Sciences. “Customers aren’t there because they followed each other into the bar. They’re attracted by the same cues: drinks, music, and the atmosphere. The same is true of mosquitoes. Rather than following the leader, the insect follows the signals and happens to arrive at the same spot as the others. They’re good copies of each other.”
Visual Cues and CO2 Create a Strong Attraction
The researchers conducted three experiments that adjusted visual targets and carbon dioxide levels. In the first test, a black sphere attracted mosquitoes only when they were already flying toward it. After reaching the object, they typically did not remain nearby and quickly moved on.
When the black target was replaced with a white object and carbon dioxide was introduced, mosquitoes could locate the source, but only at close range. Hu observed them pausing briefly, as if doing a “double take,” before gathering around it.
When both a black object and CO2 were present together, the effect was much stronger. Mosquitoes gathered in large numbers, remained in the area, and attempted to feed.
“Previous studies had shown that visual cues and carbon dioxide attract mosquitoes. But we didn’t know how they put those cues together to determine where to fly,” said Christopher Zuo, who conducted the study as a Georgia Tech master’s student. “They’re like little robots. We just had to figure out their rules.”
Human Experiments Reveal Target Areas
After identifying the importance of stationary visual cues, Zuo tested the behavior on himself inside a mosquito chamber. He wore different outfits, including all black, all white, and combinations of both.
Standing with his arms extended, he allowed dozens of mosquitoes to circle him while cameras recorded their flight paths. The data was later analyzed at MIT to determine the most likely rules behind their movement.
The mosquitoes behaved as if Zuo were simply another object. The largest clusters formed around his head and shoulders, which are the areas this species tends to target.
Luo wore a long-sleeved sweatshirt, pants, and head covering in the chamber. He said he wasn’t bitten very often.
Interactive Model Shows Mosquito Behavior
The team’s data-driven model and interactive website demonstrate how mosquitoes turn, accelerate, and slow down in response to visual signals and CO2. Users can switch between different conditions, including color, carbon dioxide, both, or neither, and observe how up to 20 mosquitoes respond. The platform also allows users to upload their own images as targets.
Findings Could Improve Mosquito Control
The researchers believe these insights could help refine pest control strategies.
“One tactic is using suction traps that rely on steady cues, such as continuous CO2 release or constant light sources, to attract mosquitoes,” Zuo said. “Our study suggests using them intermittently, then activating suction at intervals, might be better. That’s because mosquitoes don’t tend to stick around their target when both clues aren’t used at the same time.”
Reference: “Predicting mosquito flight behavior using Bayesian dynamical systems learning” by Christopher Zuo, Chenyi Fei, Alexander E. Cohen, Soohwan Kim, Ring T. Cardé, Jörn Dunkel and David L. Hu, 18 March 2026, Science Advances.
DOI: 10.1126/sciadv.adz7063
Zuo and Hu collaborated with mechanical engineering Ph.D. candidate Soohwan Kim. Additional co-authors include MIT’s Chenyi Fei and Alexander Cohen, as well as Ring Carde of the University of California at Riverside.
Never miss a breakthrough: Join the SciTechDaily newsletter.
Follow us on Google and Google News.