Fig. 1: There was a transient memory lapse 1–3 h after training on ILD discrimination on both the trained task (ILD discrimination) and the untrained task (ITD discrimination).

a Schematic of the ILD- and ITD-discrimination tasks. b–d Learning on the trained task (ILD discrimination). b Schematics of the three regimens with testing on ILD discrimination. c Mean ILD-discrimination thresholds for naïve listeners (n = 57; open square) and for different subsets of those same listeners who were tested 30 min (n = 20; gray square), 1–3 h (n = 17; red square), or 6–10 h (n = 20; blue square) after training on ILD discrimination. Test thresholds are adjusted using the initial ILD-discrimination threshold as a covariate. d Individual ILD-discrimination thresholds from the training session (abscissa) versus the test session (ordinate) for the three trained groups (symbols as in (c); lines are best-fitting regression lines). e–g Generalization to the untrained task (ITD discrimination). e Schematics of the four regimens with testing on ITD discrimination. f Mean ITD discrimination thresholds for naïve listeners (n = 74; open triangle) and for different groups of listeners who were tested on ITD discrimination 30 min (n = 38; gray triangle), 1 h (n = 8; red triangle), or 10 h (n = 15; blue triangle) after training on ILD discrimination. g Individual ILD-discrimination thresholds from the training session (abscissa) versus ITD-discrimination thresholds from the test session (ordinate) for the three trained groups (symbols as in (f); lines are best-fitting regression lines). All thresholds are raw values. All error bars indicate ± one SEM. *p < 0.05, **p < 0.001, ***p < 0.0001.