Then, we have computed standard graph measures, such as number of edges, maximum out-degree, total network weight, maximum out-strength, efficiency and clustering coefficient [6], to characterize the networks achieved during the decision-making process. The numerical results we obtained indicate that the averages over different couples of individuals are not able to discriminate among the six possible strategies (CC, DD, TT, CD, CT, DT) due to the large standard deviations (see Figure S2 and Table S1). This means that averages and standard deviations of graph metrics computed over the 26 couples do not allow for the characterization of the typical hyper-brain network associated with a specific strategy. In fact, the actual values of link weights can vary sensibly from couple to couple, due to differences in the time-responses of each brain and in the intensity of the EEG signals coming from different subjects. This is a well-known problem in neuroimaging. Even if the neural pattern corresponding to a given external stimulus, or to the performance of a certain task, does activate similar zones in different subjects, the amplitude of the associated
