Comparison among four psychophysical procedures used to assess sound localizationa)

To determine whether listeners use similar cues in absolute judgments and discrimination of sound locations, data were collected in four tasks used to assess sound-localization ability: a pointing task and three discrimination tasks (single interval, cued single interval, and two-alternative forced choice). Four stimuli were used: broadband noise, scrambled-spectrum noise (noise with levels jittered +/-20 dB in 1/3-octave bands), frozen scrambled-spectrum noise (stimuli with the scrambled noise spectrum fixed across intervals of the two-interval tasks), and a 750-Hz pure tone. Performance was better (higher d') in the discrimination tasks relative to the pointing task. However, performance in the pointing task was brought into better alignment with that of the discrimination tasks through an analysis that estimated and removed slowly fluctuating bias from the pointer responses. Discrimination performance was better for the broadband noise and frozen scrambled-spectrum noise than for the scrambled-spectrum noise and pure tone, indicating the influence of spectral cues in the former two. The results suggest that both localization (as in the present pointing task) and source-location discrimination may be based on the perceived locations of stimuli, but that spectral cues that are not associated with perceived location may influence responses for certain types of stimuli.