On top of the analyses reported in the paper, we conducted fully Bayesian analyses of trial-level behavior using a multilevel psychometric model to combine the data from all subjects (using @paulbuerkner.com 's R library brms). Ping me if interested in the code! I think this approach is the future.