Figure 1.
Stimulus with spectro-temporal variation.
Panel A: In the spectro-temporal condition the stimulus consisted of three narrowband noises. For each noise band, ten temporal segments were presented. Independent and random level perturbations were imposed on the 3 (noise band) × 10 (segment) component levels. Panel B: temporal configuration for the mid-CF noise band. The 10 segment levels were drawn independently from a normal distribution with mean µM = 55 dB SPL and a standard deviation of 2.5 dB. With identical probability, either 0.75 dB was subtracted from or added to each segment level, in order to create “soft” and “loud” trials (see text). For the low and high noise band the same temporal configuration was used. Panel C: Spectral configuration.
Figure 2.
Stimulus presented in the spectral-weights condition.
Figure 3.
Stimuli with purely temporal variation.
Panel A: broadband noise. Panel B: low-CF noise band. Panel C: mid-CF noise band. Panel D: high-CF noise band.
Table 1.
Mean sensitivity (and its standard deviation) in the intensity identification task, in terms of the area under the ROC curve (AUC), for each condition.
Figure 4.
Average normalized relative decision weights as a function of noise band. Normalization: mean of the absolute values of the three weights equals 1.0. Filled symbols: spectral weights condition. Open symbols: spectro-temporal condition, spectral weights averaged across segments (see text). Error bars show 95% confidence intervals.
Figure 5.
Average normalized relative decision weights for the four conditions showing only temporal variation, as a function of segment number. Normalization: mean of the absolute values of the ten temporal weights equals 1.0. Blue squares: low-CF noise band. Green circles: mid-CF noise band. Red triangles: high-CF noise band. Black diamonds: broadband noise. Error bars show 95% confidence intervals.
Figure 6.
Average normalized relative decision weights for the spectro-temporal condition, as a function of segment number and noise band. Normalization: mean of the absolute values of the 30 (noise band × segment) weights equals 1.0. Blue squares: low-CF noise band. Green circles: mid-CF noise band. Red triangles: high-CF noise band. Error bars show 95% confidence intervals.
Figure 7.
Temporal weights compared between the spectro-temporal condition and the single-noise-band conditions.
Average normalized relative decision weights, as a function of segment number, condition, and noise band. Normalization: mean of the absolute values of the ten temporal weights equals 1.0 per noise band. Filled symbols: single-noise-band conditions (replotted from Figure 5). Open symbols: spectro-temporal condition (same data as in Figure 6, but different normalization). Error bars show 95% confidence intervals.
Figure 8.
Individual spectral weights: observed versus predicted.
Average normalized relative decision weights, as a function of noise band. Black circles: weights estimated from the spectro-temporal condition. Gray squares: weights predicted from the sensitivity in the single-noise- band conditions (see text).