The Architecture of Smell

Published: May 8, 2017

Cracking the code on the least understood sense

How do chemical messages rising from a piece of rotting food signal a mouse to turn left to find it? How does the shape and movement of that mouse’s odor plume alert a hungry predator that it’s getting close? And precisely what goes on inside that predator’s brain when, in response to what it smells, it changes course?

These are not the questions John Crimaldi, an engineer who specializes in fluid mechanics, imagined he would be asking someday.


“All brain evolution has taken place in the form of chemical sensing, or olfaction. It is thought to be the most primal pathway in understanding brain evolution.”

- John Crimaldi, fluid mechanics engineer


“I have to admit: To lead a national neuroscience project was certainly not something I ever would have seen myself doing,” said Crimaldi, a professor in the Department of Civil, Environmental and Architectural Engineering.

But that’s exactly what he is doing. As the lead principal investigator of a $6.4 million National Science Foundation grant involving neuroscientists from six other universities, Crimaldi (the only engineer) is heading up one of the most ambitious projects yet to unravel the mysteries behind the least understood of the senses.

Scientists have long known that everything from snails to dogs uses olfactory navigation as a key survival tool, but efforts by humans to artificially mimic this skill have fallen short. As Crimaldi notes, we still use dogs to sniff out bombs and drugs.

“Even if we could build a highly sensitive artificial nose, or detector, what we have very little idea about is how to use the information embedded in the spatial and temporal structure of that odor plume to determine where the odor is coming from,” Crimaldi says. That’s where his lab comes in. Using lasers and his understanding of fluid mechanics, Crimaldi and his students are helping to develop key technological tools for virtual reality experiments at partner institutions. They create odor plumes in air or water in the lab, then use a laser to measure their precise composition, structure and movement. Those measurements are in turn used to develop a database of digital odor landscapes used to drive “olfactory generators” that, instead of projecting images, project virtual reality scents.

“Thanks to John’s group, our team has the real McCoy, measurements of real odor plumes, which we are using to create virtual environments,” said Lucia Jacobs, a co-PI from UC Berkeley who notes that because odor plumes are so complex, they are nearly impossible to simulate mathematically.

“This is an incredible and unique source of data,” she said. “No one else in the world has this. With it, I think we will be able to finally crack this problem.”

Her team will conduct experiments in humans. Others will conduct experiments in mice, having them navigate an olfactory landscape as they measure what in their neuronal networks is nudging them to go in one direction or another as odor plumes shift.

Ultimately, the information can be used to develop algorithms for programming bomb- or drug-sniffing robots. But the project has another objective.

As part of the federal BRAIN Initiative, the project also aims to use olfaction as a window into understanding the brain.

“All brain evolution has taken place in the form of chemical sensing, or olfaction,” Crimaldi said. “It is thought to be the most primal pathway in understanding brain evolution.”

Crimaldi started with a mechanical and aerospace focus before transitioning into environmental fluid mechanics. The project is an example of the interdisciplinary direction engineering has taken.

While he didn’t imagine he’d end up studying neuroscience, he’s glad he did. “It’s already opening up a whole new world of research opportunities for me and my students,” he says.