Summary: Scientists have discovered how animals differentiate between distinct odors, even those that seem remarkably similar.
While some neurons consistently identify different odors, others respond unpredictably, helping to distinguish nuanced aromas over time. This finding, inspired by previous research on fruit flies, could improve machine learning models.
By introducing variability, AI could mirror the discernment found in nature.
- Research has discovered two types of neurons: “reliable cells” that identify distinct odors and “unreliable cells” that help distinguish similar odors with experience.
- The variability in neuronal response was found to originate from a deeper circuit in the brain, suggesting that it serves an important purpose.
- This neural variability could benefit continuous learning systems in AI, making them more insightful.
Order wine at a fancy restaurant and the sommelier might describe its aroma as having notes of citrus, tropical fruit or flowers. Yet when you breathe, it can just smell… like wine. How can wine connoisseurs spot such similar scents?
Saket Navlakha, associate professor at Cold Spring Harbor Laboratory (CSHL), and Shyam Srinivasan, researcher at the Salk Institute, may have the answer. They discovered that certain neurons allow fruit flies and mice to distinguish between distinct odors.
The team also observed that with experience, another group of neurons helps the animals distinguish between very similar odors.
The study was inspired by the research of Glenn Turner, a former CSHL assistant professor. Years ago, Turner noticed something strange. When exposed to the same odor, some fruit fly neurons fired consistently while others varied from trial to trial.
At the time, many researchers considered these differences to be the result of background noise. But Navlakha and Srinivasan wondered whether the variations might be useful.
“There were two things that interested us,” says Navlakha. “Where does this variability come from? And does it serve any purpose?
To answer these questions, the team created a model of fruit fly odor. The model showed that the variability came from a deeper brain circuit than previously thought. This suggests that the variation was indeed significant.
Next, the team observed that some neurons responded differently to two very different odors, but the same way to similar odors. The researchers called these neurons trustworthy cells. This small group of cells helps flies quickly distinguish between different odors.
Another, much larger group of neurons responds unpredictably when exposed to similar odors. These neurons, which researchers call unreliable cells, could help us learn to identify specific smells in a glass of wine, for example.
“The model we developed shows that these unreliable cells are useful,” says Srinivasan. “But it takes a lot of learning to benefit from it.”
Of course, this research isn’t just for wine drinkers. Srinivasan says the findings could help explain how we learn to differentiate between similarities detected by other senses and how we make decisions based on these sensory inputs.
The results could also lead to better machine learning models. Unlike the neurons of fruit flies and mice, computers generally react the same way to the same inputs.
“Maybe you don’t want a machine learning model to represent the same input in the same way every time,” says Navlakha. “In more continuous learning systems, variability could be useful.”
This means that this research could one day help make AI more insightful and reliable.
About this latest research in olfaction and neuroscience
Author: Samuel Diamond
Contact: Samuel Diamond – CSHL
Picture: Image is credited to Neuroscience News
Original research: The results will appear in Biology PLOS