Quick intro: Barbara builds models of biological systems, often focussing on how perception shapes behavior in insects – recently covering more complex capabilities such as learning and navigation. Frequently, these models are implemented as robots – an approach we all know as Biorobotics. Barbara was the first to lay out also the theoretical foundations of this exciting new field.
You are cordially invited to attend this great talk!
Coordinates: Friday, Feb 20th, 14:00 in seminar room #005 (Takustr. 9)
Abstract: Males of the species Drosophila melanogaster – the common fruit fly – court their females. During courtship the male extends and vibrates his wing and thereby produces courtship song, which consists of two qualitatively different modes: sine and pulse song. During song, males repeatedly switch between sine and pulse. Recent studies suggest that the male switches not randomly, but chooses whether to sing sine or pulse depending on sensory input – this is despite courtship song being innate and hard-wired. Besides sensory inputs, muscle contraction is known to determine song mode and song features. I have designed simple neuromuscular models of courtship song, that allow one to study (1) the effect of neuronal frequency on song and (2) the effect of muscle contraction dynamics on song, and (3) to simulate natural looking courtship song. I have started to validate my models and simulations. My simulations excite male wild-type flies, when played back to them. Song playback is a well established paradigm to study hearing in flies and the excitatory potential of song. Further my simulations show and reproduce the link between excitation and song choice their design was based on. I expect simulated song from my models to match song parameters of real song. For matching real and simulated song I apply Approximate-Bayesian-Computation (ABC) with Sequential-Monte-Carlo (SMC) sampling to my models.
Pawel Romanczuk (MPI für Physik komplexer Systeme):
“Active Brownian Particles – From Individual to Collective Stochastic Dynamics”
Abstract: We discuss the description of biological agents in terms of coupled stochastic differential equations. This approach offers on the one hand a very flexible framwork for modeling individual, as well as large groups of interacting agents, and on the other hand allows us to perform analytical calculations on the long time/large scale behavior of such systems, which provide important insights into their general properties. Here, we will focus on particular application of these kind of models to collective motion in locusts and corresponding interaction models based on so-called “escape-pursuit” interaction behaviors.
The magazine Nautil.us has published an article on the general practise of using robots to investigate animal behavior. The article focuses on the great works of Jose Halloy and Thomas Schmickl but also mentions our work here at FU Berlin.