![](/wp-content/uploads/2021/04/abstractheadergraphic2020.gif)
Meeting Abstract
Previous behavioral studies in American cockroaches, Periplaneta americana, show they can track an odor plume even when all but a small portion of one antenna is lost. Their tracking performance improves with increasing amounts of antenna present whether bilateral symmetry is preserved or not. These observations lead us to propose an agent based model featuring spatial and temporal integration across an antenno-topic map to describe how P. americana tracks an odor plume. We focus on three models of how brains use the odor information detected by antennae. One model integrates across the antennae spatially and temporally, one compares between the two antennae bilaterally, and one compares concentration over time. We tested these models with nine different antennal configurations consisting of three arrangements (two bilaterally symmetric antennae, one antenna on the left, or one on the right) and with three different antennal lengths (10 mm, 20 mm, and 40mm). We modeled the plume using a computational fluid dynamic model coupled to diffusive mass transport. Additionally, electroantennogram and photo-ionization detector recordings provided validation of the plume structure. The integrative model is relatively inefficient (search paths are long), but robust against any type of missing antenna segments. Whereas, the bilateral comparison modelis more successful (higher fraction finding the source) but fails with a loss of one of the two antennae. Meanwhile, the purely temporal comparison model is remarkably robust, showing high levels of success in all but the highest noise trials.