Leslie Vosshall is stalking a mosquito. We’re in a small room that’s full of the little blood-suckers, which are being deliberately bred and raised. Most are safely trapped behind plastic and muslin but one plucky insect has escaped, and we’re stuck on this side of the room’s heavy door until we can find it.

It’s not easy for a human to find a mosquito that doesn’t want to be found, but a mosquito can locate us quite easily. It’s a human-seeking machine, sculpted by evolution to track the warmth of our bodies, the carbon dioxide in our breath, the smelly chemicals dissipating from our skin, and even our appearance. Vosshall has spent eight years deciphering how these creatures process these cues, in a bid to befuddle their senses and create a better generation of insect repellents. It’s a quest that has been beset by surprises and failures, which have left her with a newfound appreciation for these annoying insects—even the one that, finally, she finds and kills.

“Isn’t it beautiful?” she says, holding up its tiger-striped carcass on her blue-painted nail. It’s flattened and contorted in an almost-cartoonish way. In far better condition is the second loose mosquito that flies past our faces. “Jesus,” says Vosshall. “Let’s get out of here.”

We adjourn to her office, which feels not unlike a MOMA gift shop, full of bold prints, neon colors, and funky objets d’art. One white wall has been covered in orange and green Sharpie scribbling. Vosshall herself is wearing a white, brown, and orange 1960s-print dress; she is both chic and low-key, fiercely intelligent and self-effacing. I started working on mosquitoes in earnest in 2008,” she tells me. “It was a real pain in the ass. Getting it working was unbelievably hard.”

Before that, Vosshall studied Drosophila fruit flies—easier to work with, and not prone to biting back. In 1999, she discovered a set of proteins called odorant receptors that flies use to smell. These are built from several sub-units, some of which bind to specific molecules. But one of these units, Orco, is a universal component, as Vosshall showed in 2004. “It’s like the essential Lego piece,” she explains. “If you want to make a receptor that’s tuned to a bitter almond odor, you take an almond-sensing sub-unit and snap it onto Orco. And in Drosophila, if you take Orco away, they basically don’t smell anything.”

Getting bored of studying flies, Vosshall decided to switch to mosquitoes—a change that meant radically retooling her laboratory at Rockefeller University and learning a wealth of new techniques. “When I hit my mid-40s, I thought that I didn’t want to be an old woman still working on Drosophila,” she says, having just turned 50. She wanted to work on something with clearer applications, something that would help people.

Her plan was simple. She had shown that Orco-less flies can’t smell, and that this master protein is found in almost all insects, including the mosquitoes that spread malaria, dengue fever, and other horrendous diseases. If she could find a chemical that blocked Orco, she could develop a super-repellent, which would stop mosquitoes from finding us, biting us, and loading us with viruses. “It was such an appealing idea and it was easy to interest the Bill and Melinda Gates Foundation,” she says.

Then: “Have I developed a new insect repellent? No. Did I try? Yes.”

Her team screened a huge panel of chemicals for molecules that block Orco, and found one that did the trick. But all their efforts went up in smoke when postdoc Matthew DeGennaro created a mutant mosquito that can’t make Orco. “We flew it in our tests and it was insanely attracted to humans,” says Vosshall, laughing. To add insult to injury, the mutant mosquitoes could no longer smell DEET, our current best repellent. The Orco-blocking strategy would not only fail, but would negate our most effective mosquito-deterring chemical. D’oh.

“There was a horrible moment when we all realized that all this Gates money had gone into this project,” says Vosshall. “I feel guilty that we didn’t make a repellent, but we tried our best!”

What went wrong? “Narrowly focusing on a single sensory pathway to stop mosquitoes is doomed,” she says, because they track their prey with so many different cues. “It’s an incredibly smart strategy. From the mosquito’s point of view, there are a lot of unreliable signals in nature, so they integrate multiple pathways. They have a Plan B at every point.”

For example, mosquitoes can track humans by the carbon dioxide in our exhalations. We can’t help making the gas, and they can’t help reacting to it: add carbon dioxide into a cage full of mosquitoes, and they’ll go berserk. When Vosshall’s postdoc Conor McMeniman created a mutant mosquito that was insensitive to CO2, he reasoned that they might struggle to find prey. But when the team released these mutants in a greenhouse, complete with fake porch and makeshift garden, they had no trouble homing in on the one human in the enclosure—McMeniman himself. “We debilitated the mosquitoes and they couldn’t tell if Conor was breathing, but within three minutes, half of them found him,” says Vosshall. “The results kinda suck.”

She has since changed strategy. Her team is now systematically studying which cues are important, how they are detected, and how they collide in a mosquito’s brain to direct its flight. Mosquitoes have a Plan B? Fine. Vosshall will map it out, and Plans C, D, and E while she’s at it.

McMeniman’s work revealed that carbon dioxide makes mosquitoes more sensitive to other stimuli like heat and odors. Graduate fellow Roman Corfas showed that mosquitoes sense body heat using a protein called TRPA1. (By coincidence, so do rattlesnakes and pythons.) Former postdoc Carolyn McBride showed that the disease-carrying Aedes aegypti mosquito evolved to prefer humans over other animals by tweaking a receptor that detects sulcatone, a yeasty chemical found in human body odor. And student Molly Liu is studying the visual cues that mosquitoes are attracted to. “She makes these mag-lev mosquitoes by gluing little pieces of iron to them and suspending them by magnets,” says Vosshall. They float in this arena of LEDs, where she can display images and moving pixels.”

Vosshall is still convinced that she can make a good repellent, but she’ll need to target many of a mosquito’s senses at once. After all, that’s what DEET probably does.

Developed by the U.S. Department of Agriculture in 1944, DEET was used to protect troops in tropical countries, and licensed for public use in 1957. But despite its long history, no one really knows how it works—only that it does. Vosshall originally suspected that DEET blocks Orco but now thinks that the chemical bamboozles their sense of smell in more complicated ways. “Some olfactory pathways are unchanged, some are massively turned up, some are transformed,” she says. “It’s like completely rewiring and de-tuning a piano.”

If she can duplicate this effect, she hopes to find substances that are equally effective at lower doses, safer for infants, and longer-lasting. But only, she suspects, after more thoroughly understanding her enemy. “I’ve realized we didn’t know enough about the mosquito,” says Vosshall. “We shouldn’t be playing around with making repellents yet. We need to figure out what they’re doing before we figure out how to intervene.”