We present a study of behaviors, belonging to the domain of emotions, inquiry, persuasion and the use of information and that are deployed in time by teams of students immersed in learning processes. These are transformed into time series on which non-linear dynamics techniques are applied, obtaining the connectivity for each team. From this value and the quotient Positivity / Negativity (referred to the emotions) can be characterized learning dynamics and its performance. The Lyapunov coefficients, calculated to the time series, make it possible to obtain the Kolmogorov entropy. From this value, associated to the experimental time series, the complexity of the system and the loss of information are deduced. This allows us to construct, for the learning process, a quantitative estimate of sustainability.