\Nonetheless, behavioral observations show that cockroaches are able to accurately sense surrounding odors, and then subsequently follow resource gradients towards areas with higher amounts of food. As all ORNs that bear the same receptor, along the entire length of the antenna, converge to a single output neuron, spatial information of odour organisation along the antenna seem to be lost, creating an “information bottleneck”. Thus the prevailing hypothesis is that the animal can only determine the direction of resource gradients by continuously moving and re-sampling average concentration levels [Keene and Waddell, 2007]. There are many outstanding questions regarding this: Why would a system evolve an architecture that seems to discard environmentally relevant information? How are movement and active sensing strategies used to enhance gradient detection?
We have recently obtained neural recording data that suggests that despite of the convergent neural architecture, the time-dependent stimulus response of both the ORNs and the output neurons may be tuned to enable local gradient detection within ecologically relevant temporal and spatial scales. Strong odorants elicit stronger and faster neuronal responses, and that stimuli originating on antennal portions closer to the head (e.g. 5 mm) take less time to reach the antennal lobe with respect to more distal stimulations (e.g. 45 mm). Hence, gradients with opposite directions (stronger ahead or stronger behind) could be integrated in different ways, thus allowing the cockroach to encode information about the spatial distribution of an odorant.
To understand the mechanisms for spatial odor coding, we propose a theoretical-experimental collaborative project to develop a computational model of the odor detection system of the cockroach. Work on a model began as part of a course project earlier this year, co-advised by Jacob Davidson and Einat Couzin-Fuchs, with the course “Computational modeling in neuroscience and systems biology”. The funding will support a computer science master’s student as a HIWI to further develop the model, perform simulations that represent common odor detection scenarios and movement patterns of the cockroach, and to incorporate newly acquired data as part of the input structure of the model. Jacob Davidson will lead theoretical development and simulations, and Einat Couzin-Fuchs will lead experimental analysis and advise on incorporating experimental findings into the model.