Table 1.
Training regimen (mode and stimulus class), testing conditions and final number of participants in each of the 10 study groups.
Fig 1.
Pre- and Post-test thresholds for each training stimulus, stimulus class and testing mode.
(A) Pre-test (empty symbols) and post-test (full symbols) thresholds on the speech discrimination task for each phoneme minimal-pair stimulus. Thresholds are plotted as the phoneme number (on the continuum between 1 and 96), tested on the blocked condition (top panels) and random mode (bottom panels). Five training groups (shown from left to right on the x-axis of each panel) were tested on each mode: Blocked (B) and Random (R) Speech, Blocked (B) and Random (R) Duration, and untrained Controls (C). Note that the top and bottom panels denote different training groups (10 groups in total; see Methods for details). All error bars denote 95% confidence intervals. (B) Pre- and post-test thresholds on the duration discrimination task for each reference stimulus duration, plotted as per-cent of the reference duration. Top panels show performance on the blocked tests, while the bottom panels show performance on the random test for the same groups as (A).
Fig 2.
Learning on each stimulus class and testing mode.
Learning is shown as the difference between pre- and post-test thresholds. Individual z-scores were calculated with respect to the aggregate pre-test data across all participants tested on each stimulus class and mode (see Methods), and are thus plotted in units of standard deviation from pre-test scores (dashed line at 0) for each group. Data are shown for Blocked (A) and Random (B) Speech, and Blocked (C) and Random (D) Duration, in the groups trained on Blocked and Random Speech, Blocked and Random Duration, and untrained Controls (denoted on the x-axis of each panel). The grey box in each panel indicates the group for which training and testing was on the same regimen. Asterisks demarcate significant learning compared to the Control group (after correction for multiple comparisons, see Table 2). Error bars denote 95% confidence intervals.
Table 2.
Pre- to post-test learning vs. untrained controls in Z-score units.
Fig 3.
Group mean learning curves for all training regimens.
Group mean learning curves for the groups trained on Blocked (red) and Random (blue) Speech (A) and Duration (B). Dotted and full lines are the regression lines fitted to the mean thresholds averaged across stimuli and participants for each training session, in the groups tested on the random and blocked mode, respectively. Error bars are not shown because they overlap substantially.
Table 3.
Mean group slopes of the regression lines fitted to individual learning curves [with 95% confidence intervals].