beyond statistical significance implications of network structure on neuronal activity除了网络结构对神经元活动的统计显著性的影响.pdfVIP

Perspective Beyond Statistical Significance: Implications of Network Structure on Neuronal Activity Ioannis Vlachos*, Ad Aertsen* , Arvind Kumar* Bernstein Center Freiburg, Neurobiology Biophysics, Faculty of Biology, University of Freiburg, Freiburg, Germany Abstract: It is a common and analysis reveals statistically significant statistical tests can be applied to demon- good practice in experimental sci- modulations of the recorded neuronal strate that the emergence of correlations ences to assess the statistical sig- activity [2], then it is assumed that these among these neurons, during specific nificance of measured outcomes. spatio-temporal activity patterns are likely epochs, is indeed higher than expected by For this, the probability of obtain- to play a role for processing and compu- chance. However, this particular ‘‘Gedank- ing the actual results is estimated tation in the network. enexperiment’’ enables us to go beyond under the assumption of an appro- However, the methods used to identify merely establishing statistical significance of priately chosen null-hypothesis. If and measure the statistical significance of the activity modulations, by actually esti- this probability is smaller than these patterns do actually not justify any mating the impact of these events on the some threshold, the results are claim regarding their impact on network brain area under consideration. deemed statistically significant dynamics or function. That is, statistical