Testing and estimation for variable single evoked brain potentials

The task of validly quantifying variations in electrical activity recorded when the brain is processing external or internal events is addressed. Three new statistical tests sensitive to different types of response changes, which test the null hypothesis that there is a homogeneous signal, are presented. In the case of latency jitter, the testing procedure was developed together with a procedure for estimating the unknown signal shifts. The tests provide a statistically valid and powerful tool in investigating signal variation, even with strong colored noise. Moreover, the differential sensitivity to different types of variations allows a study of the underlying signal variability even though the single signal cannot be estimated.<>