Summary: We identify smells in the blink of an eye—often faster than we realize. A new study reveals that the heavy lifting of odor identification happens in a small fraction of a second within the olfactory bulb, not the cerebral cortex as previously believed.
By using a process the researchers call “temporal filtering,” the brain uses the very first nerve signals triggered in the first 50 milliseconds of a sniff to identify a smell while simultaneously blocking out background “noise.”
Key Facts
- Sniff Cycles: Mice sniff rapidly (250–500ms), while humans take 1–3 seconds per sniff. Despite the difference in timing, the critical identification happens at the very start of the cycle.
- Precision Optogenetics: The team used a custom-built circuit-mapping microscope to track and manipulate individual nerve signals with pulses of light (optogenetics), allowing them to “see” the exact millisecond a signal was sent.
- AI Applications: The principle of temporal filtering—prioritizing early data and ignoring late-arriving noise—could be used to make Artificial Intelligence tools process large sensory datasets more efficiently.
- Retinal Parallel: This mirrors recent findings in vision science, where it was discovered that the retina does significant “pre-processing” to identify objects before signals reach the visual cortex.
Source: NYU Langone
Mice make use of rapid nerve cell interactions in the brain’s smell center to distinguish one odor from another, a new study shows. Both mice and humans can rapidly identify odors, researchers say, in a small fraction of a second.
Led by researchers at NYU Langone Health, the study shows that the key steps involved in identifying smells happen in the mouse olfactory bulb, a part of the brain located behind the nose. The function was previously thought to occur in the cerebral cortex, a larger part of the brain known for its role in perception, awareness, and thought.
Publishing in the journal Nature Neuroscience online April 14, the study shows that a subset of nerve signals activated first and within milliseconds, when a mouse just begins to take a sniff, determine which odor is identified. The entire sniff cycle in mice can last between a quarter and a half second; while in humans, the sniff cycle is longer and takes between one to three seconds (one second is 1,000 milliseconds.)
The findings centered on the processing of signals produced by millions of olfactory sensory neurons, cells in the mouse nose that are linked to olfactory bulb glomeruli (clusters of nerve endings.) These are in turn connected to batches of mitral and tufted cells (MTCs).
The study authors found that the olfactory bulb glomeruli-MTC signals triggered within the first 50 milliseconds of the sniff cycle determined the type of odor that tested mice perceived. In the process of rapid neural computations for odor sensing that the researchers have termed “temporal filtering,” transmission of the first sets of activated olfactory nerve signals both determine the odor being smelled and block out later signals.
Specifically, the team found that the same pattern of linked glomeruli-MTC signals became active first for the same smell regardless of concentration of that odor. Once this pattern was set, activation by background odors of other sets of glomeruli blocked following nerve signals from passing along. Together, this enabled transmission of only the first set of signals belonging to the first identified smell.
“Our findings call into question a fundamental understanding about mammalian sensory processing, which is that these brain computations mostly occur in the cortex,” said study co-senior investigator Dmitry Rinberg, PhD. “The work also demonstrates for the first time how mice, but possibly humans as well, use temporal filtering to distinguish between odors.” Dr. Rinberg is a professor of neuroscience at NYU Grossman School of Medicine.
“This research is key to understanding how our sense of smell works but also how our complex neural networks are connected, and possibly how other complex biological and computational systems work,” said study co-senior investigator Shy Shoham, PhD.
Dr. Shoham, director of the Tech4Health Institute at NYU Langone Health and a professor in the neuroscience and ophthalmology departments at NYU Grossman School of Medicine, said the team’s research raises fundamental questions about the role of the cortex in processing sensory information, noting that recent advances in understanding vision similarly showed that the neural cues in the retina help tell objects apart, before any signals ever reach the cortex.
Temporal filtering, he noted, could also have application to artificial intelligence tools, if used to speed up processing of large amounts of sensory information.
Rinberg said the team next plans to examine how temporal filtering patterns in the olfactory system help distinguish between similar smells, such as citrus (lemon and orange), as well as distinguish among other sweet smells, such as those of berries or stone fruit.
The team’s latest analysis was made possible by precision optogenetics, a technique that allows researchers to specifically activate or shut down neurons using pulses of light, and to determine which individual or close-knit neurons are electrically firing when exposed to different smells.
Study lead investigator Mursel Karadas, PhD, led development of the new circuit-mapping microscope used in the study. The technique allowed researchers to stimulate and track individual nerve signals in the thin, outermost layers of the olfactory bulb.
Funding support for this study was provided by National Institutes of Health grants U19NS107464, U19NS112953, and R01DC022320.
Other NYU Langone researchers involved in the study are co-investigators Jonathan Gill and Sebastian Ceballo.
Key Questions Answered:
A: Your olfactory bulb knows it’s coffee in 50ms, but it takes a few hundred more milliseconds for that signal to travel to your cortex, where your conscious mind puts a label on it. The “perception” is instant; the “awareness” takes a moment to catch up.
A: Not necessarily. The study shows the brain is optimized to catch the first burst of data. A sharp, quick sniff (like a mouse does) provides a clear 50ms window of information. Repeated rapid sniffing helps “refresh” that window, allowing the brain to filter out background smells more effectively.
A: Yes! Current AI often struggles with “sensory overload”, trying to process every bit of data at once. By teaching AI to use “temporal filtering” (prioritizing the first 50ms of data and ignore the rest), we could create much faster and more accurate environmental sensors.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this olfaction and neuroscience research news
Author: David March
Source: NYU Langone
Contact: David March – NYU Langone
Image: The image is credited to Neuroscience News
Original Research: Open access.
“Rapid temporal processing in the olfactorybulb underlies concentration-invariant odoridentification and signal decorrelation” by Mursel Karadas, Jonathan V. Gill, Sebastian Ceballo, Shy Shoham & Dmitry Rinberg. Nature Neuroscience
DOI:10.1038/s41593-026-02250-y
Abstract
Rapid temporal processing in the olfactorybulb underlies concentration-invariant odoridentification and signal decorrelation
In a dynamic environment, sensory systems must filter out irrelevant information to construct a stable percept. Animals who rely on smell need to identify and discriminate odors despite fluctuations in concentration, yet odor receptor activation is strongly concentration dependent.
Here we explored how odor signals are transformed within the mouse olfactory bulb (OB) by developing an all-optical approach to identify the connectivity between odor receptor channels (glomeruli) and the mitral and tufted cells (MTCs), while monitoring their odor responses.
We found that the glomeruli and MTCs activated earliest in a sniff robustly represented odor identity across concentrations, whereas MTCs connected to later activated glomeruli were concentration dependent.
Furthermore, probing the responsiveness of MTCs to glomerular input found a short temporal window of excitability at a sniff’s onset, followed by prolonged odor-evoked inhibition.
Our findings demonstrate, in awake animals, that the OB implements a rapid temporal filter, which is responsible for stabilizing identity across concentrations while decorrelating responses between odors.
