Refine
Year of publication
- 2002 (1) (remove)
Document Type
- diplomthesis (1)
Language
- English (1)
Has Fulltext
- yes (1)
Is part of the Bibliography
- no (1)
Institute
- Mathematik (1)
The synchronization of neuronal firing activity is considered an important mechanism in cortical information processing. The tendency of multiple neurons to synchronize their joint firing activity can be investigated with the 'unitary event' analysis (GrĂ¼n, 1996). This method is based on the nullhypothesis of independent Bernoulli processes and can therefore not tell whether coincidences observed between more than two processes can be considered "genuine" higher- order coincidences or whether they might be caused by coincidences of lower order that coincide by chance ("chance coincidences"). In order to distinguish between genuine and chance coincidences, a parametric model of independent interaction processes (MIIP) is presented. In the framework of this model, Maximum-Likelihood estimates are derived for the firing rates of n single processes and for the rates with which genuine higher order correlations occur. The asymptotic normality of these estimates is used to derive their asymptotic variance and in order to investigate whether higher order coincidences can be considered genuine or whether they can be explained by chance coincidences. The empirical test power of this procedure for n=2 and n=3 processes and for finite analysis windows is derived with simulations and compared to the asymptotic values. Finally, the model is extended in order to allow for the analysis of correlations that are caused by jittered coincidences.