In the analysis of the response latencies in the IAT and in the impulsive avoidance tendency task, we removed error trials (5% of the responses for the IAT and 4% of the responses for the impulsive avoidance tendency task) and latencies larger than three standard deviations (0.6% responses for the IAT and 0.6% responses for the impulsive avoidance tendency task). Following the recommendations of Greenwald et al. (1998), we computed a difference score between the congruent and the incongruent trials. Similarly, in the impulsive avoidance tendency task, we subtracted the mean response latency of the avoidance trials from the mean response latency of the approach trials separately for each group, so that higher values indicate faster avoidance movements