Fig. 3
From: Bayesian nonparametric models characterize instantaneous strategies in a competitive dynamic game

Sensitivity metric reveals information about player strategies. a Observed sensitivity to opponent actions in both opponent conditions, for Participant 12. Shaded regions indicate 95% credible intervals. Blue line and shaded region correspond to the human opponent condition, green to the computer opponent. b Differences in opponent sensitivity across participants. Scatter plot of relative opponent sensitivity for each participant in the dataset for early (horizontal) and late (vertical) phases of the trial. Values are differences in log sensitivity to each opponent, averaged across timepoints and trials. Positive values indicate higher sensitivity (either early or late) to the human opponent; negative values indicate higher sensitivity to the computer. Most participants (71%) demonstrate consistently higher sensitivity to one opponent throughout the trial (denoted by red and purple points)