You are here: Home Partners Visiting Scientists George Gerstein at the Bernstein Center Freiburg
Document Actions

George Gerstein at the Bernstein Center Freiburg

George Gerstein at the Bernstein Center Freiburg


gerstein.jpgDuring April and May 2010 I had the pleasure of a second research visit to the Bernstein Center Freiburg. As in my first visit 4 years ago, this was an opportunity to catch up with old friends and colleagues Ad Aertsen, Stefan Rotter, and Uli Egert, as well as to meet and discuss with new ones.

My research project was an ongoing effort to develop, improve and calibrate tools for the detection of Synfire Chain signatures in experimental recordings. This work had begun in my previous visit to the BCCN Freiburg and resulted in a paper with Sven Schrader, Sonja Grün and Markus Diesman (Schrader et al. J Neurophysiol 2008). That work was based on the intersection of neuron identities observed at two different times.

The concept of Synfire Chain organization (a chain of neuronal groups with feed-forward connections) came originally from Moshe Abeles (1991), and immediately became popular among theoreticians. There is a large literature from many theoretical neuroscience laboratories, including the BCCN Freiburg, with investigations of chain properties like propagation, stability, interaction between different chains, representation, and compositionality.

Unfortunately, the only published evidence for Synfire Chains in real nervous systems depends on inference from the detection of repeating spike patterns in recordings of multiple spike trains. Indeed, synfire chain activity will generate repeating spike patterns, but repeating spike patterns also occur in a set of surrogate independent spike trains. If the patterns do come from synfire chain activity we would expect temporal structure and interleaving to be defined by the parameters of the synfire chains. Accordingly I examined the temporal structure of repeating spike patterns using simulated synfire data provided by Sven Schrader and Arvind Kumar. As expected, when the data come from a simulation with only one synfire chain with little unrelated neural activity, the detected patterns are structured appropriately. However if there are many synfire chains, in the simulation or if the variance in the synfire chain is high, or if there is much background activity, the temporal structures blur. Therefore, the time structure approach is not generally a reliable approach to detecting synfire activity in real recordings, and we are left with only the activity intersection method as both reliable and sensitive. In the few experimental data sets so far examined there has been no suggestion of synfire activity, but we are obtaining access to a number of other multi-neuron recordings and will actively pursue the issue of whether synfire chains occur in biology.

The social and cultural aspects my stay in Freiburg were delightful. I was able to attend a good many concerts, particularly a series of organ recitals and various interesting performances at the Hochschule für Musik. And the Augustiner Museum provided a fascinating look at medieval art and architecture. I do hope to return for another research visit to the Bernstein Center Freiburg in the not too distant future.

George Gerstein
Department of Neuroscience
University of Pennsylvania


Personal tools