In rats, olfactory bulb neurons use simple “linear summation” to interpret fluctuating odor cues from their environment.
Our sense of smell can produce powerful, immediate reactions. Encountering a foul scent—spoiled milk or a skunk—often triggers an instinctive recoil and quick action: discarding the milk or moving away from the source. The nose itself only detects molecules; the brain must transform those molecular signals into meaningful information about odor identity, intensity and location.
New research from teams at Cold Spring Harbor Laboratory (CSHL) and the National Centre for Biological Sciences (NCBS), Bangalore, shows that a core olfactory computation in rats is remarkably simple. Published in Nature Neuroscience, the study found that two principal output neuron types in the rat olfactory bulb—mitral cells and tufted cells—solve a challenging sensory problem by adding inputs in a linear way. This “linear summation” is an unexpectedly straightforward operation compared with the complex, nonlinear processes often assumed to dominate mammalian brain circuits.
Over five years and across two continents, the researchers used rats—whose olfactory systems share many features with ours—to measure how the olfactory bulb encodes odors that arrive in turbulent, fluctuating plumes. Odor plumes disperse irregularly through space and time, and an animal’s own breathing pattern further changes how odor samples reach receptors: sometimes brief, rapid sniffs, sometimes long, deep inhalations. These two sources of variability make reliable odor identification a demanding computational task.
To isolate the computation performed by mitral and tufted cells, the team controlled one key variable: respiration. By stabilizing the rats’ breathing during some trials and allowing natural, irregular sniffing in others, they compared neuronal responses under constant sampling and natural sampling conditions. The expectation—based on extensive literature showing inhibition and complex interactions in the olfactory bulb—was that mitral and tufted cells would exhibit nonlinear behavior, because inhibitory networks can dampen signals and introduce nonlinearity to a circuit’s output.
Contrary to expectations, experiments in tightly controlled conditions produced clear linear responses. As CSHL Assistant Professor Florin Albeanu explained, exposing a rat to very short snippets of an odor produced a characteristic response in a given mitral or tufted cell. Once that short-stimulus response was known, the researchers could predict that same cell’s response to longer or differently shaped odor plumes simply by scaling the short response by stimulus duration—essentially identity signature multiplied by duration.
The key property enabling this linear computation is a cell-specific latency: a consistent, brief delay before a cell either fires or is suppressed in response to an odor. That latency is stable across different sniffing patterns—rapid, shallow sniffs or slow, deep breaths—and it differs across odor identities. Each cell rapidly encodes an odor’s identity in tens of thousandths of a second via its latency profile. After identity is established by this fast, linear process, downstream olfactory cortex circuits and the olfactory bulb together extract additional attributes such as intensity, direction and distance.
This simple rule—linear summation of odor time profiles—helps explain why rodents are so efficient at smell-dependent behaviors. Rats rely heavily on olfaction for locating food, avoiding danger and finding mates, so a robust, reliable encoding strategy that tolerates variable plume structure and sniffing behavior is highly advantageous. Discovering that mitral and tufted cells implement a linear sum makes the encoding process both easier to analyze and more understandable from an engineering perspective.
The finding also challenges a prevailing assumption in neurobiology: that mammalian sensory processing is necessarily dominated by nonlinear computations. Instead, at least for primary olfactory processing in rats, a simple linear operation appears to underlie accurate, rapid odor identification despite noisy, fluctuating sensory inputs.
The research reported here received support from the Department of Biotechnology, India; a Whitehall Foundation Fellowship; CAEN Fellowship; Cold Spring Harbor Laboratory; and the National Centre for Biological Sciences, Bangalore.
Contact: Peter Tarr – Cold Spring Harbor Laboratory
Source: Cold Spring Harbor Laboratory press release
Image Source: The image is credited to Albeanu Lab, CSHL and is adapted from the press release
Original Research: Abstract for “Olfactory bulb coding of odors, mixtures and sniffs is a linear sum of odor time profiles” by Priyanka Gupta, Dinu F. Albeanu and Upinder S. Bhalla in Nature Neuroscience. Published online January 12, 2014. doi:10.1038/nn.3913