Fig. 1: Tasks and distributional regularities.

a The tone-in-noise detection task involved two phases: adaptive threshold estimation followed by the tone-in-noise detection task. Threshold estimation trials began with continuous noise and a fixation cross (750 ms), after which a 1000-Hz tone was presented with equal probability in one of three 250-ms detection windows (250 ms ISI), each indicated by a number (1, 2, or 3) on the screen. A prompt appearing 250-ms after the third detection window elicited participants’ report of the interval containing a tone. Tone intensity followed the 3-down, 1-up procedure to estimate 79% accuracy (see Methods). The noise continued through the tone-in-noise detection task, shown in the right of (a). For each trial, 500 ms preceded a 250 ms fixation cross and another 500 ms period. A 250-ms sinewave tone with intensity + 0.75 dB above the threshold estimated in the adaptive thresholding task appeared in one of two 250-ms intervals (250 ms ISI), indicated by a “1” and a “2” on the screen, respectively. Participants reported which interval contained the tone. Tone frequency varied according to the distributions in (c). b In the duration-decision task, each trial involved a 1000-ms fixation followed by a 50 or 90 ms sine wave tone (equal probability) and participants reported “long” or “short” with a button press. c Probability distributions for each experiment, as a function of acoustic frequency. Blue distributions indicate tone-in-noise detection experiments. Orange distributions indicate duration-decision experiments.