In some ways, humans take the sense of smell for granted — like knowing when dinner’s on the stove — or find it a bother, like when the family dog digs into the garbage for chicken bones.
We know that the sense of smell, or olfaction, is essential to our health and safety, but we actually don’t know that much about it.
Leading the University of Pittsburgh arm of a multi-institution effort to explore olfaction by studying animal scent-detecting abilities are Nathan Urban, medical school professor and associate director of Pitt’s Brain Institute, and noted mathematician Bard Ermentrout. Seventeen scientists from around the country will lead three collaborative projects, funded with more than $15 million from the National Science Foundation.
Among the long-term goals are developing a manmade smelling device.
“The sense of smell is something that is mysterious,” neuroscientist Urban explained. There are many aspects that we still don’t understand, he said, so it’s not surprising that we haven’t made a device that can perform smelling tasks.
Such devices could detect chemicals in the environment and smuggled material, or help with tracking fugitives or lost children, he said.
“Nature has figured this out, that goes way beyond our ability to understand. … The project is focused on bringing people together, neuroscientists, mathematicians, evolutionary biologists, to tackle how animals can do this.”
Pitt’s project will both quantify how odors move through the environment and measures how animals behave with odors surrounding them. Also involved are researchers at the University of Colorado, University of California Berkeley, Weill Cornell Medical College, New York University Medical Center and Yale University.
Each has its research to build on: Pitt and Yale have mice experiments, Berkeley uses humans and dogs, New York researchers use different species of fruit flies.
Strategies of an animal’s behavior to find food or a mate, for example, include what it does when it first encounters a smell and what it does to track it and follow it to the source.
In the first year, investigators hope to standardize ways to deliver chemicals to the animals, so they can be compared in a careful way, Mr. Urban said. They also plan to develop techniques to measure concentrations of chemicals in the local environment, to see patterns in how they move there and to then be able to do well-controlled experiments.
In the early stages of the behavioral studies, he said, “we will be able to identify what discrete part of the brain is involved in the task.”
The goal is to then develop a device that, for example, could find the source of hazardous chemical leaks. Now the solution for some hazards is particularly low-tech. As an example, Mr. Urban said, the safest land-mine detection system appears to be the use of trained Gambian pouched rats, known as HeroRats.
“It’s remarkable to me that is the best solution we have is to train rats.”
Animal scent-detecting abilities are vast, he said.
“There are tens of thousands of chemicals in our environment that we are sensitive to… Humans have over 400 odor receptor genes; dogs have 800 or so and mice and rats have 1,000, or 1,200.” Each gene encodes a sensor for a different class of chemicals, he added.
He explained how trained mice in Pittsburgh provide data:
Researchers watch mice from below a transparent plexiglass table as the mice follow trails made by scented crayons.
“The animal is trained that if they follow the trail, they get a reward. We can record the activity of neurons in their brain as they perform this task … we need to know if the odor concentration is going up, or if it is going down,” he said, explaining that when the scent is stronger, it’s noted which area of the brain is stimulated.
Never a man to take smells for granted — as an expert cook, he had been able to taste the ingredients of any new dish he encountered — mathematician Ermentrout now suffers a partial loss of the sense of smell caused by a bad cold several months ago. As he waits for the sense to return, he is diving into the research and said math can help describe the sequence of mouse behavior.
“We know that they sniff, and as they sniff, they move their heads, sampling and then sampling again — to know which way to turn. Then we start to design algorithms, to make virtual mice on a virtual trail. Then we want to ask how they do with fluctuation. How do they fail ... how can they tell the odor?”
Among other questions: “How does the mouse sniff and how does it hold [the smell] in its memory? What neurons are involved?”
A smelling device, Mr. Urban said, would need a sensor that could detect a range of chemicals and be able to determine the intensity of a chemical. And, like most organisms, he said, it would have to mimic behavior — like sniffing or otherwise creating air flow — that would sample the air, analyze a scent’s concentration and compare that sample with an earlier one.
“Ultimately this would be used by a robotic device, that could do this by itself.”
Jill Daly: jdaly@post-gazette.com or 412-263-1596.
First Published: September 29, 2015, 4:00 a.m.
