Continuous data-stream computation by a minibrain


State of the art, own contribution: Large brains obtain vast processing power by sheer number of neurons, but insect brains are constrained to operate with much smaller circuits. For specific tasks, however, they achieve excellent performances although they cannot use large-scale population coding to average over internal noise sources. Hearing in grasshoppers and crickets demonstrates how small processing modules extract relevant information with millisecond precision in real time from a continuous input stream. In contrast to cortical circuits, the systems performance can be quantified by psychophysical tests and electrophysiological recordings from identifiable neurons.



Objectives: A crucial problem of all neural systems are stochastic components and these insects form a prime example in which one can study how a constrained system has evolved to optimize the signal-to-noise ratio in its neural representations. Using psychophysical and electrophysiological approaches, we will study (i) sound localization and relate spike-train variability of individually identified interneurons to the reliability of behavioural responses within one individual; and (ii) computational principles of extracting spatio-temporal information in real time which depend on a delicate balance between excitation and inhibition in single neurons.



Description of work: Multielectrode recordings will be used to assess the variability as well as covariances of neuronal activity where we will profit from the theoretical competence of S. Grün. This shall be complemented by simultaneous intracellular recordings from two neurons, to determine connectivities of neurons and to explore their variability on a finer scale of synaptic potentials. As a next step the properties of neurons will be characterized in patch clamp studies (input from the Heinemann and Schmitz labs) and potentially also imaging studies (input from the Menzel lab). Subsequent pharmacological manipulation will reveal biophysical properties of single neurons. Together with the known and newly determined connectivities this study will foster realistic modelling of the auditory network. Psychophysical studies of individual animals with subsequent electrophysiological tests will provide crucial information about the influence of variability and noise on real-time performance.


Principal Investigators
Hennig, Matthias Prof. Dr. rer. nat. (Details) (Physiology of Behaviour)

Duration of Project
Start date: 09/2004
End date: 12/2010

Last updated on 2020-10-03 at 16:46