Animals balance efficiency and flexibility during natural behavior to make their decisions in different contexts. In a recent study published in Current Biology (PDF) the Murthy lab has developed a novel behavioral paradigm in an experimentally accessible laboratory model (the mouse) that sheds light on how animals use alternate strategies depending on experience.
Mice have a highly developed sense of smell which they use to gather information about the world around them. How odor cues are used by mice to navigate to sources is not well understood. This issue is particularly difficult when considering airborne odors, since they are primarily dispersed by turbulent odor transport. This is appreciated readily by anyone watching a steaming cup of coffee or smoke blowing from a chimney. It has not even been clear whether mice can use airborne odor cues (as opposed to surface-bound odors) to navigate. David Gire (now an Assistant Professor at the University of Washington) and others in the Murthy lab placed mice in an arena in which an odor was released from one of several locations. Mice had an incentive to get to the odor source as quickly as possible because they would get greater reward. Watching mice navigate to odor sources over many trials and sessions, the investigators were able to make several inferences.
First, they were able to show that mice that were relatively new to the arena navigated to the odor source using airborne cues. To show this required a physical model of odor transport, since there are no easy methods to directly measure the spatiotemporal distribution of odors. Agnese Seminara in SEAS at Harvard (now at the CNRS in Nice, France) created computational fluid dynamics models based upon the flow conditions in the experimental arena using numerical simulations of the well-known Navier-Stokes equations describing fluid flow. Overlaying experimentally-observed mouse trajectories on the simulated distribution of odors, they were able to infer the likely temporal sequence of odor concentrations encountered by mice as they navigated to the odor source. They found that an algorithm that allows a searcher to move up a locally-measured odor gradient is sufficient to explain the trajectories of mice to the different odor sources. Such a simple gradient climb algorithm fails at distances larger than about 40 cm from the source because turbulent flow creates highly fluctuating, intermittent odor encounters for the searcher.
A second important result of the study was that mice that were experienced with the geography of the arena learn to quickly move among the reward sites, ignoring odor cues until they are very close to the source. This strategy was better than the earlier one that involved navigation using odor plumes, since mice were able to increase the average reward obtained. Together, this work demonstrates that while mice can dynamically track complex sensory cues, they are also capable of adaptively using learned geographic information from their environment to more quickly and efficiently navigate to rewards. Ongoing research will use this new paradigm to study how neural circuits integrate memory and sensory cues to control the selection of behavioral strategies.