Computer modeling of neural networks complements physical studies of odor discrimination
Collaboration Linked to the Sweet Smell of Success
Innovation is often linked to interaction. When the Jerome L. Greene Science Center is completed, researchers and their labs from different areas of the Columbia neuroscience community will be housed together in a single facility, fostering increased collaboration and exciting new discoveries.
Which is not to say that collaboration doesn’t happen every day at Columbia and especially at the Kavli Institute for Brain Science (KIBS), because it does–often in interesting ways.
For example, Larry Abbott, PhD, William Bloor Professor of Neuroscience and Co-Director of the Center for Theoretical Neuroscience, has been collaborating with Richard Axel, MD, University Professor at Columbia and Investigator at the Howard Hughes Medical Institute. According to Dr. Abbott, this came about because of an inquisitive student.
“Students often make interesting links because they are exposed to different ideas and disciplines every day,” he says. “In this case, the student specifically wanted to work with both of us because we each are able to provide a piece of the puzzle he wanted to try to solve.”
Picking up the scent
The student, then in a combined MD/PhD program in medicine and computational neuroscience, was working on a dissertation focusing on theoretical models of olfactory discrimination in the fruit fly Drosophilia.
Dr. Axel’s studies revolve around flies and mice, as their olfaction systems are very similar. But something very unusual happens as they process smells, according to Dr. Abbott.
“In the initial stage in the olfactory system,” he says, “information about odors is transferred from receptor neurons to the first ‘relay station’, otherwise known as projection neurons. This connection is very precise. From there, however, things get more disorganized. The projection neurons fire off axons to the brain in what appears to be a very disordered fashion.
“The question is why is this so and what does it mean?”
A sense for survival
All sensory functions in a living organism process information needed for safety survival, as well as sustenance. Identifying odors is important, not just for finding suitable food and mates, but also for steering clear of poisons and other dangers. But the ability to discriminate among odors is only half the task. There have to be appropriate associated behaviors, such as drinking or running away. Some of these behaviors are innate and some are learned.
“Given that the wiring of information about odors from the projection neurons to the brain is disorganized,” says Dr. Abbott, “we wanted to know if different individuals might process and react to the information in different ways–and would that give some individuals an advantage over others?”
Dr. Axel’s lab is able to physically manipulate the biological system, directly stimulating the neurons that respond to odors. For example, if a researcher touches a neuron associated with drinking, the animal will start licking on a spout. Dr. Abbott’s piece of the puzzle was to conduct a mathematical analysis to see if there are properties of odors that are carried through even if the wiring is random.
“It turns out there is such a thing and you can use mathematics to describe how the properties can be preserved. By knowing which aspects are preserved or not preserved,” notes Dr. Abbott, “we’re able to predict behaviors.”
For Dr. Abbott, the opportunity to collaborate on biological experiments is welcome. “Most experiments require imaging neurons that you can’t do with a live human,” he says. “What I like about theoretical neuroscience is that we can measure anything and change anything we want. In general, though, I like to keep my models on the simple side; I resist putting in a lot of unnecessary detail. I want to know what are the minimal set of features that a neural circuit would have to have in order to explain the phenomenon.”
Ideally, cross-disciplinary collaboration creates its own repeating cycle, where findings on one side lead to explorations on the other side that lead to new approaches and new findings and on and on. Dr. Abbott notes that by analyzing his models, he can identify variables that can inform future experiments.
“I can give Dr. Axel or any other collaborator a few new things to measure that might make a useful difference,” he says. “Or I might ask the investigator to manipulate things in a certain way because of what the model suggests, and he or she will ask me for different kinds of models based on what the experiment revealed.
“By going back and forth between the constraints of experimental reality and the freedom of modeling work,” Dr. Abbott continues, “we can usually discover things we wouldn’t have been able to find if we were working on our own.”