Figure 2.
Initial model comparison and filter optimization. The graph shows predictive validity (i.e., ability to distinguish CS+ from CS‐ responses) quantified as Akaike information criterion (AIC, smaller is better). Dark gray: initial filter settings; light gray: optimized filter settings. Dashed lines represent the decision thresholds with respect to the best model (absolute AIC difference > 3). a: Experiment 2. b: Experiment 3. c: Experiment 4.