The Bernstein Center for Computational Neuroscience Freiburg

Announcement for the next
Informal Seminar
Benjamin Staude
Unit of Statistical Neuroscience, Theoretical Neuroscience Group
RIKEN Brain Science Institute, Wako, Japan

Identification of assembly activity in massively parallel spike trains

Friday, June 20th, 2008
14:00 h TIME
Seminar Room 1048, Institute for Biology I, Hauptstrasse 1
Hebb's cell assembly hypothesis postulates dynamically interacting groups of neurons as the building blocks of cortical information processing. Synchronized spiking across large neuronal groups was later suggested as a potential signature of active assemblies. The computational advantages of the resulting correlation-based assembly codes over population rate codes attract both computational and experimental researchers, however compelling experimental evidence in favor of the assembly hypothesis has been notoriously difficult to provide. Recent technical developments allow the spiking activity of large neuronal populations (~ 100 neurons) to be recorded simultaneously. Methods for assembly detection in such massively parallel spike trains typically test the number of occurrences of precisely timed spike patterns against the assumption of independence. However, the number of different patterns to be tested in these approaches grows exponentially with the number of recorded neurons, requiring unfeasibly large samples, and rendering their application to spike trains recorded in physiological experiments practically impossible. As a consequence, most attempts to detect active cell assemblies resort to pairwise interactions. These, however, do not permit the inference of large synchronized neuronal pools, and are insensitive to sparsely occurring synchronous events. Furthermore, a clear statistical distinction between the signature of assembly activity and population rate codes has not been provided, which impedes the unambiguous interpretation of measured correlations. Taken together, the limited experimental support for the cell assembly hypothesis can be assigned, among other things, to a lack of suitable analysis tools.
Here, we provide a theoretical framework to overcome these limitations. Firstly, we provide a model-free distinction between the signatures of cell-assemblies (synchrony) and population rate codes (rate covariance) within the framework of doubly stochastic point processes. The relevance of the resulting distinction is illustrated by analyzing simulated data sets. Secondly, we present a novel procedure to detect synchronized spiking in large neuronal pools that circumvents the need for vast sample sizes. Based on the filtered compound Poisson process as a parametric model for the superimposed and discretely sampled activity of recorded neurons, we devise a statistical test for the presence of synchronized neuronal groups in a recorded population. When applied to simulated data, the test is surprisingly sensitive for higher-order synchrony present in the data, even if their effects on pairwise correlation coefficients are very small (in the range of ~ 0.01). The applicability of the method is illustrated by estimates for the required sample size and the robustness against deviations from the Poisson assumption.
The talk is open to the public. Guests are cordially invited!