Supplemental Material for Cognitive Training Reduces the Strength of Pavlovian Biases

These discs were associated with small (1) and large (4) virtual rewards for correct responses (the association between disc size and reward magnitude was counterbalanced across participants).This acquisition phase comprised 20 trials, 10 of each trial type, presented in a random order.In the second phase, participants were told that they would now see discs of intermediate sizes as well, and had to categorise these as small or large too, depending on which group they seemed closest in size to.In fact, the intermediate, ambiguous discs were all exactly halfway between the original stimuli (i.e.AE 300px.On half of the trials, the ambiguous stimuli were rewarded as if they were small, and in the other half, they were rewarded as if they were large.There were 120 trials in this phase, 40 trials each of high, low and intermediate stimuli, presented in a random order.The main outcome was the proportion of trials on which participants treated the intermediate discs as if they belonged to the more highly rewarded group, quantifying a positive affective bias.Risk Taking Task (Rutledge et al., 2016) In this economic decision-making task, participants had to make a series of choices between a guaranteed outcome or a risky 50-50 gamble.On gain trials, the certain outcome was a gain of between 20 and 60 points, and the gamble returned either 0 points or a larger gain (between 1.7 and 5 times the guaranteed outcome); loss trials were equivalent but with negative values for both the guaranteed outcome and the gamble; and in mixed trials, the guaranteed outcome was zero, and the gamble outcomes were either a gain of between 30 and 150 points, or a loss of 0.2 to 2 times the gain.There were 150 trials in total, 50 gain trials, 50 loss trials and 50 mixed trials, presented in random order.