# MIT and Georgia Tech Crack Mosquito Flight Code, Revealing Three Distinct Attack Patterns

*2026-03-23 12:02 UTC · Science · 1 min read*

> Scientists built the first 3D model of mosquito flight, finding three distinct attack patterns based on sight and smell cues.

Scientists from MIT and Georgia Tech have built the first 3D mathematical model of how mosquitoes hunt humans, analyzing over 477,000 flight paths from 53 million data points. The study, published in Science Advances, identifies three distinct behaviors: a rapid "fly-by" dive when mosquitoes only see a target, a slow hovering "double-take" when they only detect CO2, and a focused "orbiting" pattern when both visual and chemical signals align — the critical moment before a bite.

Led by MIT mathematician Jörn Dunkel, the team found that combining sight and smell triggers the most dangerous behavior. Mosquitoes that can both see and smell a person switch to steady circling, locking onto their target like a predator. This explains why dark clothing and heavy breathing make you a prime target — you are giving mosquitoes both signals at once.

The findings have direct applications for mosquito trap design. Current traps typically use only CO2 lures, but the research shows that adding visual cues could dramatically increase capture rates, potentially reducing malaria and dengue transmission.

## Sources

- [news.mit.edu](https://news.mit.edu/2026/new-model-predicts-how-mosquitoes-will-fly-0318)
- [sciencedaily.com](https://www.sciencedaily.com/releases/2026/03/260322020247.htm)
- [earth.com](https://www.earth.com/news/mosquito-flight-patterns-reveal-how-they-track-humans/)

---
TechTools News · https://techtools.cz/tools/news/article/993-mit-and-georgia-tech-crack-mosquito-flight-code-revealing-three-distinct-attack
Česká verze: https://techtools.cz/tools/news/clanek/1028-mit-a-georgia-tech-rozlustili-kod-letu-komaru-a-odhalili-tri-odlisne-vzorce
