Fast and dynamic odor encoding in rat olfactory cortex during odor discrimination task

Abstract

Event Abstract Back to Event Fast and dynamic odor encoding in rat olfactory cortex during odor discrimination task Active movement of sensory organs plays an important role in many sensory modalities. In olfaction, animals sniffing frequency shifts dynamically depending on animals? behavioral states or task demands. Previous behavioral studies from our group showed that rats can achieve highly accurate odor discrimination with a single sniff, suggesting that sniffing is a unit of odor encoding (Uchida and Mainen, 2003). However, very little is known about how active sniffing affects odor encoding by neurons, because most neurophysiological studies in mammalian olfaction have been performed in anesthetized animals. In this study, we wished to address how odor representations develop over sniff cycles and how these representations relate to behavioral performance. We focused on the olfactory cortex because it is closer to behavioral output than the olfactory bulb and therefore we expected that it may be easier to establish direct links between behavioral performance and neuronal activity. We recorded neural activity from individual cells in the anterior piriform cortex of rats using a multi-electrode recording technique while they performed a two-alternative odor discrimination task. The rats were trained to sample an odor at a central odor port and to go left or right to a choice port to obtain water rewards. In each session, at least six odors were used, with half of them assigned to either the left or right reward port. Simultaneously, respiration was monitored using a temperature sensor implanted in the nostril. Olfactory cortical neurons showed robust and transient odor responses with specific latencies that were tightly locked to sniff cycles. Many neurons showed odor-specific adaptation or facilitation at the second sniff. We then examined how the ensemble activity evolved across sniffs. Using principal component analysis, it could be seen that ensemble activity followed odor specific trajectories. Interestingly, the ensemble activity in the first sniff was significantly different from that in the second sniff. This contrasts with the highly similar trajectories repeated over multiple sniff cycles in previous studies in anesthetized animals. We next quantified information content within each sniff cycle by asking how well an ideal observer can discriminate six odors based on the neural activity. To do this we used several different pattern classification methods including support vector machines. This analysis showed that the discriminability of odors developed very rapidly within the first sniff and that the second sniff provided less information. Using the combined spikes during the first and second sniffs did not improve odor discriminability significantly compared to using only spike during the first sniff. In summary, an informative odor representation in the olfactory cortex of behaving rats appears to develop very rapidly within a single sniff and is not substantially augmented odor representation in the second sniff. These observations may help to explain the behavioral observation that multiple sniffs in the odor sampling port does not result in increased discrimination performance. This study suggests that olfactory processing during active sniffing favors rapid processing over prolonged temporal integration during an odor discrimination task.