Predictable sound exposure increases behavioral spontaneous frequency generalization but reduces trial-to-trial variability.

<p>(A) Scheme of a single PPI trial: startle noise was preceded by a prepulse tone with a Δf of between 0% and 50% below the background tone of 16 kHz. (B) Normalized PPI as a function of frequency change between the prepulse and the background tone of 16 kHz for control, predictable, and random groups. Continuous line indicates a fitted logistic function (ANOVA, group F_2,189 = 13.59, <i>p</i> < 0.01; corrected pair comparisons: <i>p</i> < 0.001 predictable versus control and <i>p</i> < 0.01 predictable versus random). Dash line: discrimination threshold. Control <i>n</i> = 7; predictable <i>n</i> = 8; random <i>n</i> = 9. (C) Mean AUROCCs calculated from the behavioral data in B. Continuous line indicates a fitted logistic function (ANOVA, group F_2,189 = 1.3, <i>p</i> = 0.27. Control <i>n</i> = 7; predictable <i>n</i> = 8; random <i>n</i> = 9). (D) Scheme of a single PPI trial with a tone background of 11.3 kHz. (E) Normalized PPI as a function of frequency change between the prepulse and the background tone of 11.3 kHz for control and predictable groups. Continuous line indicates a fitted logistic function (ANOVA, group F_2,81 = 18.92, <i>p</i> < 0.0001; group × frequency interaction F_2,81 = 3.06, <i>p</i> < 0.01). Control <i>n</i> = 4; predictable <i>n</i> = 7. (F) Mean AUROCCs calculated from the behavioral data in E. Continuous line indicates a fitted logistic function (ANOVA, group F_2,81 = 10.36, <i>p</i> = 0.0019). Control <i>n</i> = 4; predictable <i>n</i> = 7. Error bars represent SEM. Numerical data for this figure found in <a href="http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.2005114#pbio.2005114.s001" target="_blank">S1 Data</a>. AUROCC, area under ROC curve; PPI, prepulse inhibition of the auditory startle reflex.