Fall Course on Computational Neuroscience in Göttingen

As in the last years there will again be a fall course on computational neuroscience hosted by the BCCN Göttingen in 2009. It is ment as a tutorial for BCCN students as well as an open fall course for external students.

• Michele Giugliano: In vivo-like, noisy stimulation paradigms in in vitro cellular electrophysiology
• Robert Gütig: (working title) Spike-timing based sensory coding and processing
• Jason Kerr: Population imaging in vivo: from the awake to the anesthetized
• Benjamin Lindner: Interspike interval statistics and response properties of neurons in the fluctuation driven regime
• Gianluigi Mongillo: Collective dynamics of recurrent networks with short-term synaptic plasticity.
  
  

Abstracts

In the last few years, an increasing amount of experimental papers employed noisy waveforms as an in vitro stimulation paradigm for cellular electrophysiology.
The idea of probing a dynamical system with non-deterministic inputs is quite common in system theory. Noise may be employed as a practical alternative to the estimation of the (impulse) input-output transform properties of the system. Similarly, psychophysics and experimental (e.g. visual) system neurosciences introduced similar estimation techniques, inspired more or less by the same consideration or by information theoretical approaches. Therefore, noise has been employed as a general purpose input, maximizing information content at the initial processing stage, and acting as a statistical representative input of larger variety of (natural) input stimuli.
More recently, researchers started employing current-clamp or dynamic-clamp stimulation by fluctuating inputs, in in vitro experiments involving cellular and synaptic electrophysiology. Although meeting tough criticism by reviewers, these authors were inspired by theoretical approaches, proposing that noise may quantitatively mimic in vitro what a neuron would experience under realistic condition in the intact brain of an animal.
Background synaptic activity, irregular firing, non-selective responsiveness, synaptic convergence and recurrent connectivity are all typical hypotheses to account for a relatively noisy environment (e.g. cortical) neurons are embedded in vivo.
In this lectures, I will introduce some of the hypotheses that inspired such a new impulse in experimental electrophysiology and present a selection of results from both system and cellular levels. I will also focus on experimental work I have been directly involved in and attempt at providing a comprehensive picture of what can be learned from these stimulation paradigms.

Many neurons show unmistakable stochastic features in the generation of action potentials (spikes) due to various sources of noise and irregularity in their dynamics. In order to elucidate the role of these fluctuations for different aspects of neural activity (spontaneous spiking, response to external stimuli, network behavior), a number of researchers have studied simplified neuron models (one or two-dimensional dynamical systems with dynamic noise and an additional
weak driving) in order to gain some intuition about the possible behavior in the stochastic setting. In my talk I introduce a number of basic measures to quantify the stochastic activity, e.g. the
time-dependent firing rate, the coefficient of variation, the interspike interval (ISI) density, spike train power spectra, and the input-output coherence. Next, I discuss for some simple models
(neurons with threshold noise or white noise driving) how to calculate measures of spontaneous spiking (e.g. rate and CV) and of the response (e.g. the susceptibility) in the framework of the Fokker-Planck equation and the first-passage-time formalism. For models involving
more time-scales (of adaptation variables or of finite noise correlations), the calculation and the functional relevance of ISI correlations is highlighted. In the final part, I explain how network
behavior can be quantified based on single-neuron properties (spontaneous power spectra and response functions).

Short-term synaptic plasticity (STP) profoundly affects dynamics in recurrent cortical-like networks, resulting in the appearance of new activity patterns. This lecture intends to present a collection of phenomenological models and of analytical techniques which allows for the study of collective network dynamics in presence of STP. After a whirlwind tour of the physiology of chemical synaptic transmission, I will introduce three phenomenological (i.e., low-dimensional) models of short-term synaptic plasticity. While being a somehow rude simplification of the underlying biochemical processes, these models still capture well quantitative aspects of STP patterns as exposed by in-vitro experiments. At the same time, as they involve only few variables, they are amenable to analytical treatment. I will then show how, with these models, to compute the characteristics of the post-synaptic response as a function of the statistics of synaptic activation. The analytical results derived will be used to investigate collective network dynamics of simplified rate models which describe population activity, and subsequently of more realistic networks of spiking (integrate-and-fire) neurons. I will focus on the following issues: (i) How the number of stationary states of network dynamics depends on STP patterns (e.g., depressing or facilitating); (ii) The stability of stationary states as a function of STP patterns and of the spiking statistics (for networks of integrate-and-fire elements); (iii) The emergence of non-trivial temporal dynamics for population activity (e.g., population spikes), and their functional implications, when asynchronous steady states destabilize.

Aims

The course is intended to provide graduate students and young researchers from all parts of neuroscience with working knowledge of theoretical and computational methods in neuroscience and to acquaint them with recent developments in this field.

The speakers provide an overview on important aspects and recent developments in their fields of expertise by means of three-hour tutorials. In addition to the tutorials, participants will gather in small groups and study one out of a number of recent research papers that are related to the respective tutorial. The "self-study" will be supervised by the speakers such that (based on the introduction given in the tutorials) a profound insight in the main ideas can be obtained. The results of the self-study will be shared with other participants and discussed with them and the speakers during the presentations.

The particular form of the course has proven successful in previous courses (1999 and 2001 at Bochum, 2003, 2004, 2005, 2006, 2007 and 2008 at Göttingen). It combines lecturing with an active interaction with the main ideas of the topical fields in a way which has proven efficient given the time constraints of the course.

Set-Up

One of the main objectives of the course will be to enable participants from any field of neuroscience to study recent research papers on their own. Each day of the course is devoted to a different topic. Part of the teaching will be in form of a tutorial, but there shall be room for the activities of the participants. In previous years each day has been divided into three phases:

1. Before noon there will be a tutorial consisting of two lectures (2 times 90 min) where (as a rule) the first one should introduce the topic of research and the second one more specifically should provide background information for the study of a number of recent key papers in the field.

2. Each paper is then assigned to a group consisting of about four participants. The papers shall be scanned already before the course (the papers can be downloaded some weeks before the course). In the early afternoon, participants discuss the paper and prepare a presentation to the members of the other groups. Speakers will be available during the self-study phase to answer questions related to the papers.

3. Later in the afternoon there are slots of 15 min presentations of these papers by one or more representatives of each group and discussions together with the tutorial speakers and the members of other groups.

In the evening there will be opportunities to participate in various social activities.

Schedule

	Mon. 21.	Tue. 22.	Wed. 23.	Thu. 24.	Fri. 25.
8:45-10:15	registration	Lindner	Giugliano	Gütig	Mongillo
coffee break
10:30-12:00	Kerr	Lindner	Giugliano	Gütig	preparation
lunch break
13:00-14:30	Kerr	preparation	preparation	preparation	presentation
coffee break
14:45-16:15	preparation	presentation	presentation	presentation	Closing Discussion
16:30-18:00	presentation	city tour	Grenzlandmuseum	Mongillo

Questions
Can be asked via our contact form

Former Fall Courses on Computational Neuroscience in Göttingen

2008 2007 2006 2005 2004 2003