This 6th G-Node Course on Neural Data Analysis comprises four course modules, one module for each day. Each module treats a specific topic that will be presented by one faculty member. Each module includes an introductory lecture plus a total of six hours of practical course work. All students are required to prepare in advance on the basis of the introductory reading lists below.
Alex Loebel (Ludwig-Maximilians-Universität München)
Synaptic responses to a series of pre-synaptic action-potentials are complex and non-linear. This phenomenon, termed short-term synaptic plasticity, is observed at all stages of the nervous system, from synaptic connections at the spinal cord to neocortical connections. Some connection types exhibit short-term depression, in which the responses diminish with the stimulus series; while other types exhibit short-term facilitation, in which the responses facilitates. In this module we will learn how to model such responses, and how to extract the models' parameters from experimental data. Using the Jackknife statistical method we will evaluate the accuracy of our fits. Time permitting, we will also discuss the stochastic version of these models, with which the quantal parameters of synapses can be estimated.
Matlab data files for the different exercises
Matlab script file, for exercise (1i)
Matlab script file, for exercise (2ii)
Matlab script file, for exercise (2iii)
Thomas Wachtler (Ludwig-Maximilians-Universität München),
Martin Nawrot (Freie Universität Berlin)
This course module introduces the students to the analysis of directional tuning in single unit recordings from the motor cortex of the behaving monkey. Goal of the module is to understand basic concepts of neuronal tuning curves and rate coding, and to practice the analysis of single-unit spike train data. Students will analyze single unit activity recorded from the monkey primary motor cortex during the preparation and execution of reaching movements.
Fabian Sinz (Eberhard-Karls-Universität Tübingen)
Originally developed in the 1940ies for applications in communication, information theory has become a standard tool in computational neuroscience where it is, for instance, used to quantify the available information about a stimulus in responses of neurons, or to extract those features in the stimulus that carry most information about the neural response. In the course, we will look at simple scheme how to estimate a lower bound on the information between stimulus and the corresponding neural response which uses the tight link between information and the notion of predictablity. We will apply this method to recordings from macaque V1 neurons from the paper by Berens et al. listed in the references below.
Jan Grewe (Eberhard-Karls-Universität Tübingen)
In this module spectral methods like power-spectrum estimation, reverse reconstruction and the coherence will be introduced. We will apply these methods to data recorded in electroreceptors of weakly electric fish to reconstruct the stimulus from the responses and to estimate the relation between the stimulus and the neuronal response.