Harvard University - Department of Molecular & Cellular Biology

SNIFFING OUT THE ODOR CODE: A ROLE FOR SPIKE TIMING

by Kevin Cury and Naoshige Uchida

November 3rd, 2010


(L to R) Naoshige Uchida and Kevin M. Cury

How would you describe the smell of fresh baked bread, a rose, or an ocean breeze? The world of smell is extremely diverse, and with intangible qualities that make it difficult to put into words. However, animals exhibit a remarkable capacity to identify and respond to odors in just a fraction of a second. In fact, a study by Uchida and Mainen (2003) demonstrated that a single sniff cycle, lasting about 150 milliseconds, is sufficient for high performance odor identification. How does the brain accomplish this feat?

To begin with, one must consider the basic unit of communication utilized by the nervous system -- the action potential, or spike. In a simple case, a neuron may indicate the presence of a particular odorant, say menthol, by increasing the rate at which it fires spikes. Alternatively, the language, or "code", that the brain uses to represent odors may instead be comprised of the specific timing or patterning of spikes. Such a “temporal code" has attracted many researchers as it may allow the brain to represent a larger repertoire of stimuli and in a shorter amount of time.

In our study published in Neuron, we provide three main lines of evidence for the existence of an inhalation-coupled temporal code in the olfactory bulb -- the first odor processing center and the exclusive recipient of output from the nose. First, we examined the responses of principal neurons in the olfactory bulb of rats while they engaged in an odor discrimination task. Our analysis revealed that a population of these neurons could accurately encode the identity of an odor within just 100 milliseconds of inhalation when their spike timing was evaluated at a resolution of tens of milliseconds. As a second measure, we took advantage of variability in the animals' discrimination time to identify the relevant features of the neural response: the fine-scale timing of spikes, but not the total spike count over the entire sniff cycle, was predictive of how quickly the rats would make up their mind about the odor's identity.

For our third and final measure, we compared the neural response of the same set of principal neurons across two distinct modes of odor sampling -- alert, high frequency sniffing and passive, slow breathing. Features of the response that are conserved across sampling frequency could be used by the brain to recognize odors regardless of the behavioral context. Again, our results favored spike timing, as only the fine-scale temporal features of the response were preserved between sampling modes. All together, our study supports a spike-timing based code as the most likely candidate for the brain's representation of odors within a single sniff at the earliest stage of olfactory processing.

Read more in Neuron

View Nao Uchida's Faculty Profile