Credit: Stacey Brown

Timing is a thorny issue for the chemical senses. Principal neurons in the vertebrate olfactory bulb and insect antennal lobe have dynamic odor-evoked responses that can long outlast odorant exposure. The temporal pattern of these responses is thought to be important for distinguishing different odorants, but in a natural environment, odor stimuli have their own temporal structure. With an animal's movements and the intermittent arrival of odors in the air, new odors are likely to interrupt responses to previous ones. Experience tells us that a second sniff of a rose still smells like a rose, but how does the olfactory system sort out these potentially conflicting time courses?

On page 1568 of this issue, Mark Stopfer and colleagues address this question in the olfactory system of the locust. They exposed adult locusts to trains of brief odorant pulses with natural interpulse intervals, and made intracellular and extracellular recordings from projection neurons (PNs) in the antennal lobe. In most cases, responses varied over successive pulses, showing that the temporal structure of odorant presentation did influence the temporal pattern of PN responses.

To determine the downstream consequences of this interference, the authors considered known features of the locust olfactory system. Each antennal lobe contains 830 PNs, and more than 100 PNs converge onto each of about 50,000 Kenyon cells in the mushroom body, a brain area important for olfactory memory. Odorant stimulation evokes oscillations in the antennal lobe, and Kenyon cells seem to integrate convergent PN input over 50 ms, approximately one oscillation cycle. The authors pooled 117 PN recordings from multiple experiments and constructed activity vectors over 50-ms windows as well as ensemble activity trajectories over the entire PN responses to different patterns of odor pulses.

In contrast to the varied responses in individual PNs, ensemble PN activity showed highly similar responses to multiple pulses. Response trajectories for repeated odor pulses overlapped, with successive pulses appearing to partially reset the circuitry such that each response began similarly and followed a similar trajectory, irrespective of the pulse timing. A template chosen from any 50-ms window of a response could accurately classify odorant identity for responses in different presentation patterns.

These results suggest that the problem of temporal interference may be adequately solved by converging PN input onto Kenyon cells, which also responded reliably to repeated odorant pulses. However, this solution depends on the ability of newly arriving odors to reset ensemble activity in the antennal lobe. What restricts PNs to repeatedly return to the same response trajectory despite ongoing dynamic activity? Can an ongoing response to one odorant also be reset by the arrival of a different smell? Researchers undoubtedly will continue to sniff around for these answers.