The degree of complexity of the cortical processes related to different cognitive actions should logically produce an image in the electric and magnetic signals of the human brain. Hence, those measures of chaoticity will depend on the particular task being investigated. In many cases, the duration of a brain process does not allow for the generated signal to be treated as stationary. Therefore, the application of the standard method of nonlinear system theory is often questionable. The chaoticity of the process has to characterize the predictability of future state of the system. In the stationary case, such a quantity can be directly expressed by the largest Lyapunov exponent or by K-S entropy.In this study, we performed a test for the applicability of the local Lyapunov exponent for the description of the chaoticity of the brain processes measured in EEG and MEG experiments.' We demonstrate an algorithm for computation of chaoticity based on the local Lyapunov exponent and present possible applications of this method for specific cases with a diagnosis of schizophrenia or of tinnitus. We also show that chaoticity is able to detect critical transitions (phase-transition-like phenomena) which occur in the dynamics of neural mass activity at a specific point in time.