392238 Electric fish: active sensing in biology, models and hardware implementations (S) (SoSe 2013)

In this seminar we will review two key aspects of neurosciences. The first section is dedicated to the analysis of sensory signals contaminated by reafference. The principle idea is that sensory input induced by ego-motions can hinder the ability to sense the world as it interferes with externally generate stimuli. Nervous systems must be able to differentiate between these two inputs. A strategy found throughout the animal kingdom is to route copies of movement commands or higher-order feedback signals to the brain regions processing and comparing the peripheral and central inputs. This is referred to as corollary discharge or efference copy mechanism.
We’ll discuss an extensively studied sensory system where the fundamental importance of corollary discharge mechanisms coupled to adaptive plasticity based on precise spike-timing has been studied since long: the weakly electric fish of the Mormyrid family. As a starting point the organizers will give an introductory lecture on weakly electric fish, including some “in vivo” demonstrations. Then will get an overview about corollary discharge mechanisms in the animal kingdom. This will be followed by an introduction to the network in Mormyrids, experimental studies of corollary discharge mechanisms and spike-timing dependent synaptic plasticity in this network. These biological papers serve as a base to learn how generalisations of experimentally derived principles are achieved through modelling studies. These principles can be further exploited in dedicated hardware implementations to develop novel technologies, as presented in the hardware oriented paper.

In the second section we will focus on a specific sensory capability, the ability to navigate as well as communicate using self-generated electrical signals. The model species in this second part will be members of weakly electric fishes from the Gymotiform family. Noteworthy these do share many biophysical and neuronal commonalities with the species investigated in the first section, but differ in that they lack efference copy mechanisms and need to distinguish between communication and localisation stimuli that are embedded within the same neuronal stream. In this regard they offer unique opportunities to study neuronal coding of stimuli varying in both temporal and spatial domains. Again we’ll start off from an introductory presentation to familiarize participants with the behaviour and the neuronal network being investigated in the following talks. In these talks several information theoretic approaches to the analysis of neuronal coding will be presented, with an emphasis on how neuronal performance can be quantified and how different channels of information (localisation & communication) can be achieved using a single type of receptors. These experimental and analytical studies will be followed by talks focussing on the behaviour of these fishes and how it can be modelled as well as a theoretical study on parallel coding. Finally we’ll learn how the principle of electrolocation and the biomechanics of the unique locomotion of these animals are currently being used in hardware-oriented robotics.

Participants will be requested to read beforehand all the literature presented during the seminar and present 1 or 2 papers (depending on the number of participants). A collective discussion will follow the paper presentation. Participants will also be requested to upload a short report about the presented paper on the seminar wiki page.

To access more detailed informations about the course, please login Stud.IP system (https://elearning.uni-bielefeld.de/studip/) and enroll to the "Electric fish: active sensing in biology, models and hardware implementations" course (SS 2013).