For human beings, smell is a nice if somewhat auxiliary sense. In many corners of the animal kingdom, though, it’s the essential way organisms understand the world.
“For most animals, if you lose your sense of smell, you’re done,” says John Crimaldi, who studies environmental fluid mechanics at the University of Colorado Boulder. “For us, it’s harder to have a good intuitive sense of it, because we don’t rely on that sense as heavily as some creatures.”
Crimaldi is among a group of researchers from a range of fields that has recently started work on an ambitious project to establish how olfactory perception works. The effort, which is part of the White House BRAIN Initiative, seeks to map the way organisms respond behaviorally to odors and nail down the neural pathways that undergird those responses.
“[Olfactory navigation] is something animals are really good at and we engineers are not good at,” says Katherine Nagel, a neuroscientist at the New York University School of Medicine who will be conducting experiments on flies and their sense of smell. “We don’t know how to build a robot that finds a gas leak, but if you leave a piece of fruit out on the table, flies will be there in the morning.”
The project, which got underway late last year, involves a number of different kinds of experiments. Several involve positioning mice or flies in a “virtual odor environment,” similar to the virtual reality environments generated by devices like Oculus Rift — except in these, the visual and sound stimuli are replaced with precisely controlled odors.
“Instead of visual projectors providing visual information into the right and left eye, we have olfactometers that release puffs of odorant into the right and left nostrils,” says Crimaldi. “The animal perceives something that’s entirely realistic, and we know exactly what it’s smelling because we’re providing that information.”
To create these virtual odor environments, the researchers first have to understand the physics of odor plumes. Crimaldi is doing this in his laboratory, where he releases chemical compounds into a controlled air or water environment and shines a laser on them. The laser reveals the internal spatial patterns of an odor plume and allows his lab to track how those patterns change as the plume moves away from the source and diffuses in the environment. You might imagine the plume as a stream, spreading out evenly from a center line, but the actual spatial arrangement is much more complicated — with strands of odor here, dense knots of odor there, and many empty pockets where there’s no odor at all.
“When you have this odor plume and it’s being transported by a moving body of air, that moving body of air is almost universally turbulent, and the patterns that develop in the odor plumes as you move downstream become very complex,” says Crimaldi. “It’s very intermittent and episodic, and there are lots of places where there’s no odorant and other places where there are strong, intense filaments of odor.”
Knowing the physics of odor plumes lets the researchers do two things. First, it allows them to create the virtual odor environments. Second, it allows them to begin to think about the kinds of information animals might extract from an odor plume in order to decide how to behave. Given that following an odor trail (or escaping from one, as animals do when they sense a predator) is not as simple as running on a line, the researchers want to know what animals can calculate about the location of an odor source from the complex composition of the odor plume at any given point in space.
“There are all sorts of statistics we can show that are different at this location [in the odor plume] than that location, but we don’t know that those are the statistics the animals are using,” says Crimaldi. “We’re pretty confident there’s information in there, but we can’t a priori say which ones the animals are making use of.”
As the mice and flies navigate the virtual odor environments, researchers track their behavior and try to create a general view of how animals make decisions based on olfactory stimuli.
“The idea is we try to understand how the behavior relates to each animal and the animal’s abilities and then generate a model that will predict [behavior] for other animals we haven’t tested,” says Justus Verhagen, a neuroscientist at the Yale School of Medicine and researcher on the project.
At the same time the researchers track behavior, they’ll also be observing the brain neurons that fire in conjunction with the behavior. By doing this, they’ll try to understand not only what animals do based on olfaction but also the exact neural infrastructure that makes such highly developed behavior possible.
“We want to replicate the physical neural network that actually exists in the brain,” says Crimaldi. “We want to do it the way the rat’s brain does it, not just to get the same end result, but to understand the exact mechanistic process.”
The upshot, they hope, will be to provide a clearer view — or a stronger whiff — of an important dimension of perception.