• autocorrelation
• autoregressive-model
• electroencephalogram
• intransitives
• long-memory
• task
• tool mediated;
• transitives
• upper limb

EEG is a neuroimaging technique widely used for the study of the brain dynamics involved in the execution of various tasks, such as motor ones. To this end, there are several methods used in the literature to characterize the EEG signal, among which the autoregressive models. These include univariate, multivariate and fractionally integrated models. In this work, the use of ARFI models was proposed with the aim of extracting the long memory component from the EEG signal, described by parameter d. These models have proved useful for the analysis of various time series, but have never been applied to the EEG signal. The application of these models to the EEG signal proved to be very useful as significant differences were found in the parameter d between movement and rest. This implies that this parameter contributes to the discrimination between movement and rest. Furthermore, significant differences were also found between different categories of movement. Since the results obtained are very promising, this analysis could be successfully used in applications such as robot control, brain computer interface, rehabilitation.