If we could see smells, what would they look like? That’s the question a group of scientists at the University of Colorado Boulder are currently trying to answer. Scientific American reports that fluid mechanist John Crimaldi and his team of researchers have developed a new method of visualizing aromas using a 50-foot tank, 5000 gallons of water, and some high-tech lasers.

Crimaldi and his team inject a chemical dye called a “surrogate odor” into the tank of water and shine lasers through it to see what scents look like as they move through space. As it turns out, scents look like smoke but move in flowing, tentacle-like filaments. A small tube is used as an artificial nose to visualize how our inhalations and exhalations affect the movement of the surrogate odor.

Crimaldi tells Scientific American that visualizing scents is just the beginning of a much larger project. Crimaldi and his team want to figure out how humans and animals map out scents and use their sense of smell to navigate. Eventually, Crimaldi even hopes to develop a "smellbot" that can distinguish between different scents on its own.

The goal of the project is to better understand our sense of smell, one of the more mysterious and less-researched of the senses. But it also has practical implications. Currently, we use animals to sniff out bombs, drugs, and missing people, often in incredibly dangerous situations. Crimaldi hopes the smellbot could one day replace the animals and humans risking their lives to find bombs or uncover the victims of natural disasters.

In the PBS NewsHour video “What a Smell Looks Like,” Nsikan Akpan interviews Crimaldi in his University of Colorado lab. Crimaldi shows off his smell visualizer and explains how smells move through space. While the smell visualizer is just the first step in the larger research project, Crimaldi says it has already changed the way scientists understand odor. “You can learn more about this problem from sitting and looking at something like this than just about anything else,” he explains. Check it out:

[h/t Scientific American]