Participants
Thirty-two healthy participants (13 women) with normal or corrected-to-normal vision, without a history of neurological or psychiatric diagnosis, performed the experimental task. Their age range was between 19 and 23 years old (M = 21.4, SD = 1.3). Participants were randomly assigned to one of the two experimental conditions (see experimental task below). The experimental procedures were performed in accordance with institutional guidelines and were approved by the Ethical Committee of the Pontificia Universidad Católica de Chile. All participants gave their written informed consent. All experiments were carried out at the Laboratorio de Neurociencia Social y Neuromodulación of the Centro de Investigación en Complejidad Social (neuroCICS), Universidad del Desarrollo, Santiago, Chile.
Power and sample size
To calculate the minimum sample size and the power of the current study, we used the behavioral effect as the primary outcome. A similar study related to the effect of the presence of others during socioeconomic decision making 27 shows an effect size of η2=0.149, which is a large effect 44. Taking into account publication bias, we set an intermediated effect size of η2=0.1. Thus, for the between-within factor interaction in a 2x2 mixed ANOVA with a power of (1-β)=0.95 and a significance level of α=0.05, the minimum sample size to find the expected effect was n=32.
Experimental task
The experimental task consisted of a modified version of the DG, in which participants had to choose between two possible allocations of money for themselves and for another player (hereinafter the other). Within the two possible choices, there was always one that gave participants the option of being prosocial. In the case of this task, a prosocial decision is defined as choosing the option that provides higher incomes to the other player regardless of the consequences for the participant. All participants were informed that they would receive an amount of money at the end of the experiment, which would correspond to the sum of two factors. One is composed of one of their own decisions (one trial selected at random), and the other corresponds to the decision made by a previous player, also selected at random. In the latter case, the amount corresponds to the money that a previous participant paid to the other player during his/her performance of the task. This design has the aim of highlighting that decisions have real consequences for both the participants themselves and another person.
The sample was randomly assigned to one of the two groups: the empathy group and the observer group. For both groups, participants first performed a round of the task as a control condition, in which their decisions were told to be anonymous. After this first round, participants assigned to the Empathy group performed a second round of the task under EC. The EC is the same task as in the CC, but at this time, participants were told that this part of the experiment was part of a larger study that involved donation of money to a real charity institution but that these donations did not depend on the specific decision of the participants. They were informed that, under this condition, the decisions were anonymous. It is important to note that, at this point, all the participants received the explicit clarification that the amount of money that they allocated was not a direct donation to the charity institution. No further information was given about the charity institution, and a blurred picture of a child who would benefit from this donation was shown to the participants during the rest of the experimental task. Participants assigned to the observer group, after playing the CC, played a round of the task under the OC. The OC is the same task as in the CC, but at this time, participants were observed by another researcher who is part of the research team while they were playing, so their decisions were no longer anonymous. They were informed that the observer was a sociologist who conducted the observation as part of another study. A blurred picture of the observer was shown to the participants for the rest of the experimental task (as in the EC). Moreover, a confederate playing the role of a sociologist entered the experimental room several times during the experiment, taking notes and observing the participants while playing.
The decision task consisted of the selection of one within two distributions of money that were presented in the higher and lower parts of the screen. These distributions or allocations were presented separately for 1500 ms each, as shown in Figure 1. Both allocations involved money for the participant and for another player. The other player was identified as a future participant in the experiment. The amounts for participants were presented on the left side of the screen in yellow, and the amounts to the other player were presented on the right side of the screen in blue (see figure 1). The amounts were presented for 1500 ms, and a visual cue (green fixation cross) indicated to the participants that the decision could be delivered. During all conditions, participants faced three types of choices or cases that were randomly presented: others’ advantageous inequity (OAI), others’ disadvantageous inequity (ODI) and altruistic inequity (ALT) (see Figure 2). In the ALT choices, the prosocial option involves a personal cost to the participant. In the OAI cases, there is no such conflict as in the ALT case, given that the prosocial option involves higher earnings for the other participant but no personal costs to the own players. In the ODI cases, the prosocial option prevents the other participant from obtaining lower earnings with no personal costs to the player. Finally, participants were told that sometimes the game would choose the opposite option to the one chosen by them (e.g., if the participant chose the prosocial option, then the nonprosocial option would be displayed as the chosen option). If they want to return to their original option, they will be punished in that trial by losing a fixed amount of USD$5, but if they want to keep the "error", then that choice (the opposite of the one they wanted) will be considered. This was used as a way to confirm the strength of participants’ decisions. At the end of the experimental task, all participants received a payment from one of the played trials (selected at random) plus what a previous participant left to them.
EEG
Brain activity was recorded from 64 scalp electrodes using a Brain Vision amplifier system (BrainProducts, Germany). BrainVision Recorder was used to record brain activity (electrode impedance <5 kΩ, 0.15–500 Hz, 1000 samples/s). All recorded EEG epochs were individually checked for artifacts by visual inspection. Artifacts were first automatically detected using a threshold of 150 uV and a power spectrum greater than 2 std. dev. for more than 10% of the frequency spectrum (1 to 30 Hz). Blinking was extracted from the signal by means of ICA. Trials that included artifacts detected automatically and confirmed by visual inspection of the signal were eliminated. The artifact-free EEG material was recomputed to average reference and digitally bandpassed filtered to 0.1–45 Hz. Whole power distribution was computed using Wavelet transform, with a 5-cycle Morlet wavelet, in a −1.5 to 1.5 s window around the onset of the second offer. For all analyses, we used the dB of power related to the baseline (15 seconds acquired at the beginning of each block).
Statistical Analysis
We used the Kolmogorov-Smirnoff test for normality. When the data did not meet the normal assumption, we used nonparametric tests. For the EEG statistical analysis, we first fitted a general linear model (GLM) of the power of the oscillatory activity per trial in each participant (first-level analysis, see 45–48) using the following equation:
Power(t,f) = β1 + β2ΔA1 + β3 ΔA2 +β4 ΔA1*ΔA2 + β5T+ β6ΔA1*T + β7 ΔA2*T +β8 ΔA1*ΔA2*T (1)
where β1 is the intercept, β2 is the slope (coefficient) of the variable ΔA1 (differences between the allocation for the player and the other in the first presented distribution), β3 is the slope of the difference between the allocations for the player and the other in the second presented distribution (ΔA2), and β4 is the slope of the differences between the allocation for the player and the other in both offers (note that this regressor takes values other than zero only in the ALT case). Additionally, we added a regressor for the experimental condition (T, which takes the value 1 when the decision is made during the experimental manipulation, observer or empathy conditions, and 0 in the control condition) with its respective slope (β5), together with the interactions between the experimental condition and the other regressors (ΔA1*T, ΔA2*T and ΔA1*ΔA2*T). Then, we obtained a 3D matrix of the normalized β-values estimated (electrode, time, frequency, β-value/standard error) for each regressor and participant. We then explored for differences between groups (observer and empathy) and conditions (control and experimental) using the Wilcoxon test (second-level analysis). To correct for multiple comparisons in the time-frequency domain, we used the cluster-based permutation (CBP) test 49. Briefly, in this method, the clusters of significant areas were defined by pooling neighboring sites (in the time-frequency chart and adjacent electrodes) that showed the same statistical effect (cluster threshold detection, CTD, uncorrected p < .05). The cluster-level statistics were computed as the sum of the statistics of all sites within the corresponding cluster (e.g., Z value for Wilcoxon test). We evaluated the cluster-level significance under the permutation distribution of the cluster that had the largest cluster-level statistics. The permutation distribution was obtained by randomly permuting the original data (i.e., permuting a specific regressor per trial for within-subject analyses or group labels for between-subject analyses). After each permutation, the original statistics test was computed (i.e., Wilcoxon test), and the cluster-level statistics of the largest resulting cluster were used for the permutation distribution. After 5000 permutations, the cluster-level significance for each observed cluster was estimated as the proportion of elements of the permutation distribution larger than the cluster-level statistics of the corresponding cluster.