A dog unearths a bone buried months ago. A moth flies across a forest toward an orchard. A person opens container after container in the fridge, trying to pinpoint what smells spoiled. In each case, the brain is solving a difficult problem: locating the source of an odor carried by turbulent, unpredictable air currents.
Odors do not arrive as steady streams. They travel in plumes—bursts of molecules separated by clean air. Field studies and atmospheric models have suggested that the frequency of these odor encounters changes with distance: farther away, encounters are sparse; closer, they are frequent. But whether animals can detect these differences—and how the brain computes them—has remained largely untested.
Now, researchers in the lab of MCB’s Venkatesh Murthy report that mice can categorize odor sequences based on encounter frequency and reveal how early olfactory circuits encode that information. Published in Nature Communications (PDF), the study sheds light on how the brain transforms intermittent sensory signals into useful guidance.
Teaching mice to “smell distance”
To test whether mice could distinguish sparse from frequent odor encounters, co-first authors Luis Boero, postdoctoral fellow in the Murthy lab, and Hao Wu, former graduate student in the lab, and colleagues designed a behavioral task that mimicked natural odor plumes. Laboratory mice were trained to categorize sequences of identical odor pulses delivered randomly over several seconds. Sparse pulse trains—analogous to being far from an odor source—required licking one port for a water reward; frequent pulses—suggesting proximity—required licking the other.
Because pulses were distributed randomly across the full time window, mice could not rely on just the first or last second. They had to integrate information over time.
“Field observations have told us that encounter frequency changes with distance, but we didn’t know if animals could actually perceive those differences,” Boero said. “By forcing the mice to evaluate pulses over several seconds, we could ask whether they were extracting that information in a meaningful way.”
The mice learned the task readily. Their performance showed they could categorize odor sequences based on pulse frequency, effectively using temporal structure as a proxy for distance. Computational analyses revealed that their errors scaled with the number of pulses presented—a hallmark of quantity-estimation behavior seen across species.
“They’re not just reacting to a few pulses,” Boero said. “They are integrating evidence across time.”
Breathing shapes perception
Smell is tightly linked to respiration. Odor molecules primarily enter the nose during inhalation; during exhalation, access to receptors is reduced. That means olfactory sampling is gated by the breathing cycle.
The team asked whether pulses arriving during exhalation were less influential and whether inhalation-aligned pulses carried more weight in decision-making.
By monitoring breathing during the task, the researchers aligned each odor pulse with the respiratory cycle. Pulses delivered during inhalation had significantly greater influence on the animals’ final choices than those arriving during exhalation.
To probe the neural basis of this effect, the team performed wide-field calcium imaging in the olfactory bulb, the first relay of odor information. Activity in olfactory sensory neurons peaked when pulses coincided with inhalation and declined toward exhalation. This respiratory modulation closely mirrored the behavioral weighting of pulses.
“The timing of a pulse relative to the breathing cycle determined how strongly it drove activity in the olfactory bulb,” Boero said. “What’s striking is how tightly that neural modulation correlates with the pulse’s influence on behavior.”
Rather than compensating for poorly timed pulses, the brain appears to integrate a respiration-weighted signal as it is received.
Following the signal downstream
The olfactory bulb projects to several brain regions, including the piriform cortex, which connects to frontal areas involved in decision-making. To examine how pulse information is represented there, the team conducted in vivo electrophysiological recordings from piriform neurons while mice performed the task.
Because pulses were delivered at random times, identifying which spikes corresponded to which pulses was challenging. The Murthy lab collaborated with engineers at Harvard’s John A. Paulson School of Engineering and Applied Sciences, who developed a deep-learning framework to disentangle pulse-locked neural responses.
The analysis revealed a probabilistic code. Individual pulses did not reliably activate the same neuron, nor did any single pulse engage all odor-responsive neurons. Instead, each pulse activated a subset of neurons with varying firing strength.
Some piriform neurons showed respiration-dependent responses similar to those in the olfactory bulb, but roughly half appeared insensitive to respiratory phase.
“That heterogeneity was fascinating,” Boero said. “It suggests that piriform cortex may be using different subpopulations to represent different aspects of the olfactory experience.”
To determine whether piriform cortex encoded sensory evidence or the animal’s choice, the team performed decoding analyses. By examining the activity of as few as about 20 neurons during the stimulus window, they could reliably recover whether a trial was sparse or frequent—and even predict the animal’s decision.
Despite this predictive power, piriform neurons largely returned to baseline between pulses, suggesting they primarily relay dynamic sensory information rather than finalize decisions.
Questions ahead
While the study clarifies how early olfactory circuits encode encounter frequency and how respiration shapes that code, important questions remain. Are animals effectively counting pulses, or summing odor concentration over time?
“We know they integrate information over several seconds, but whether they are counting pulses or computing a running average is still unclear,” Boero said.
He is now recording from additional brain regions to pinpoint where sensory evidence is transformed into a committed choice.
“The olfactory system always finds a way to surprise us,” Boero said. “This feels like the beginning of a much bigger story about how the brain navigates a world of intermittent signals.”
(PDF)
