The masker consisted of a lower and an upper frequency band. These bands were composed of three pure tones each, and it was placed symmetrically around the signal frequency. The spacing between the three tones in each masker band was 100 cents (1 cent  =  1/1,200 octave), while the distance between the signal and the masker components closest to it was 350 cents. The level of each masker component was 50 dB SPL. The signal was roved in frequency between 600 and 2,400 Hz. In order to investigate the effect of frequency region, we used three interleaved adaptive tracks estimating thresholds separately in a LOW (600–952 Hz), a MID (952–1,512 Hz) and a HIGH (1,512–2,400 Hz) frequency region. In each track, the level of the signal was initially set at 60 dB SPL and was then varied adaptively using a 2-down 1-up rule targeting the 70.7 % correct point on the psychometric function (Levitt 1971). Track selection was pseudorandom, with a maximum of three consecutive trials per track permitted. The step size was 4 dB for the first four turnpoints and 2 dB thereafter. A block of trials was terminated when at least 16 turnpoints per track had occurred.

Each trial began with the presentation of a copy of the signal, at 50 dB SPL, in order to eliminate uncertainty about the signal frequency. The first observation interval started 500 ms after the offset of this signal cue. There were three precursor types: SYNCH, ASYNCH and SILENT. In the SYNCH conditions, the precursor was an exact copy of the masker, except that its duration was 50 ms. It was presented five times before the test sound, with an ISI of 62.5 ms between precursor bursts. In the ASYNCH conditions, the components of each precursor burst, rather than being gated simultaneously, had a 12.5-ms onset asynchrony; there was no ISI between the precursor bursts. The order in which the six precursor components were successively gated in each ASYNCH burst was random. For both the SYNCH and the ASYNCH conditions, the time interval between the middle time point of the last precursor burst and the onset of the test sound was 256.25 ms. As a consequence, the ISI between the offset of the last precursor burst and the onset of the test sound was 231.25 ms in the SYNCH conditions and 200 ms in the ASYNCH conditions. In the SILENT conditions, the test sound was separated from the beginning of the observation interval by 793.75 ms of silence. All tones were gated with 10-ms raised-cosine ramps.

In the SYNCH and ASYNCH conditions, the precursor was presented either to the same ear as the test sound (“Ipsi”) or to the opposite ear (“Contra”), while the initial signal cue was always presented to the same ear as the test sound. Listeners completed twelve sessions. During each session, they completed one block of trials for each precursor type (SILENT, Ipsi SYNCH, Contra SYNCH, Ipsi ASYNCH and Contra ASYNCH); these five blocks were randomly ordered. The first two sessions were considered as practice sessions, and the final thresholds were computed as the arithmetic average of the remaining ten threshold estimates, for each precursor type and frequency region. The stimuli were presented via TDH-39 headphones fitted with audiocups that ensured no interaural crosstalk at the presentation levels we used.