Defecting or not defecting: how to "read" human behavior during cooperative games by EEG measurements.

Timeline of the experiment.At each round, or trial, players are asked to choose either to cooperate (C) or defect (D) through a special keyboard. A trial (k) consists of two distinct time intervals. During the first interval, players have to communicate their strategies on the base of the outcome at the previous trial (k-1). Typically, this interval ranged from 0.5 seconds to 2 seconds. After communicating their choice, a report summarizing the strategy and the score at the trial (k) is displayed for 4 seconds. At the beginning of this second interval, the two subjects make the new decision to be communicated in the next trial (k+1). In particular, we considered the first second (i.e., 1 s of EEG recordings) as period of interest (POI) for the initial decision-making processes.

Inter-brain connectivity for pure strategies in the Alpha band.Two generic players are represented by the realistic head models used to estimate the cortical activity in the same six regions of interest (ROIs). Different colored points indicate the barycenters of these ROIs on the semi-transparent cortex. For the sake of simplicity, we didn't label the ROIs of each subplot, but just two for the CC (7_L, 10_L), TT (7_R, 10_R) and DD (CMA, ACC) subplot. Only links between the two brains are illustrated in each hyper-brain network, i.e. the inter-brain connections. The size and the color of each directed connection represent the PDC values of a representative couples of subjects in the Alpha (8–13 Hz) frequency band.

Pie diagrams of efficiency E, divisibility D and modularity Q in the Theta band.Top panels: from left to right the diagrams represent the percentage of cases - over the 26 couples - in which graph efficiency E is minimal, whilst the divisibility D and modularity Q are maximal. Bottom panels: percentage of cases - over the 26 couples - in which E is maximal and D and Q are minimal. Blue areas represent pure cooperation CC, red areas represent pure defection DD, green areas represent pure tit-for-tat TT. Mixed situations CD, CT, and DT are represented by white areas. The results are reported for the Theta band (4–7 Hz). Similar pie diagrams for the other frequency bands are in Figure S3.

Scatter plot of efficiency E, divisibility D and modularity Q during cooperation (CC), defection (DD) and tit-for-tat (TT).For each couple x, and each strategy τ, the Z-scores are computed as in formula 5 (Materials and Methods section). Then <Zτ(x)> is evaluated as an average of Zτk(x) over all the 26 couples. For each strategy, and each frequency band, we report in panel (a), the average Z-score for the measure of divisibility, <Zτ(D)>, vs. the average Z-score of the efficiency, <Zτ(E)>, and in panel (b), the average Z-score of divisibility, <Zτ(D)>, vs. the average Z-score of the modularity, <Zτ(Q)>. Red squares represent DD values; blue circles represent CC values and green diamonds TT values. The Greek letter beside each symbol indicates the considered frequency band.

Average values of total strength sin+sout of the ROIs during cooperation (CC), defection (DD) and tit-for-tat (TT) in the Beta band (14–29 Hz).Values of total strength (y-axis) are obtained by considering the ROIs (x-axis) of each single subject within the couple. Thus, they represent the average of 52 subjects, i.e. 26 couples. Each line corresponds to a different task: CC (blue circles), DD (red squares) and TT (green diamonds). Vertical bars denote 0.95 confidence intervals. Single stars indicate the ROI where the DD strategy is significantly different (p<0.001) from the CC and from the TT strategy. A double star marks the ROI where all the three strategies are significantly different (p<0.001).