An experiment was conducted to further explore sound localization for
frontal sound incidence. The question of interest was whether two resp
onse measures, accuracy in loudspeaker identification and choice respo
nse time would be differentially affected by variation in three parame
ters. These were the number Of loudspeakers (three, five and seven), t
he separation between speakers (15 degrees versus 30 degrees), and sti
mulus frequency (500 versus 4000 Hz). Twelve normal-hearing subjects w
ere tested in a semi-reverberant room that modelled real-world listeni
ng. Each was presented 12 blocks of listening trials, across ,which th
e 12 conditions were presented in random order. A block comprised 20 r
andom presentations of a 300-ms one-third octave noise band from each
speaker. Subjects responded using a set of microswitches in the same c
onfiguration as the loudspeaker array. Accuracy decreased as the numbe
r of speakers increased and their separation decreased. Response time
increased with number but was unaffected by separation. The effect of
frequency was relatively small. A two-stage conceptual model, consisti
ng of sensory encoding and non-sensory decision-making, provided a rea
sonable framework for the results. (C) 1997 Elsevier Science Ltd.