Talk by Aldana M. González Montoro
- 19 Jun 2013 13:30
- 19 Jun 2013 14:30
Bootstrap methods for dependent simultaneous spike-trains
The use of bootstrap methods for hypothesis testing in the context of parallel spike trains is becoming popular because of their flexibility and the increase of computational power. In correlation analysis, for example, these methods are mostly used to assess the significance of associations among neurons and, therefore, resampling procedures are thought to destroy these associations. In this talk we will address the problem of resampling from data under a different type of null hypotheses. In particular, the problem of testing for a difference in the amount of correlation will be discussed. In this context, the distribution of the test statistics under the null hypothesis is imitated by means of bootstrap methods that preserve the existing dependence among the original spike trains. The procedures will be illustrated with real and simulated data.