Effect of aspirin on the temporal effect in simultaneous masking with on- and off-frequency maskers

International audienceThis psychoacoustical study examined individual differences in spatial hearing and aims to develop learning methods to improve auditory localization performance in virtual environments. Nowadays, acoustic cues that are extracted by the human auditory system to localize sound sources (interaural time and level differences, monaural spectral cues, distance cues...) have been well identified. Many technological tools are now designed to produce a 3-dimensional perception of sounds emitted through headphones. To create the illusion of a sound originating from outside the head, these tools use Head-Related Transfer Functions (HRTFs) that are not individualized. Hence, the "realism" of the simulation is not perfect, since these listeners "listen with someone else's ears". The study presented here is an attempt to find the best way to adapt listeners to these "distorted" 3-D auditory environments. At the first place, we examined the individual differences existing in auditory-localization capacity for real sound sources. We also looked at the type of systematic errors in localization responses as well as spatial asymmetries in these errors. To do so, we measured auditory localization responses of 22 right-handed and 10 left-handed normal-hearing subjects in an anechoic room. Twelve loudspeakers were placed behind a curtain at +/- 7, 21, 35, 49, 63 and 77° in the horizontal plane, at 2.10 meters from the listener's head. The curtain was graduated every 10° (from -90° to +90°) to help listeners to give their answers. The signal was a burst of narrow-band noise (250-2000 Hz) lasting 50 ms repeated at a rate of 10 bursts/sec, for a total duration of 500 ms and an overall level of 80 dB SPL (measured at the theoretical center of the head). Listeners had to indicate where they had heard the source, by placing a cursor on a computer screen displaying a picture of the curtain with its graduations. They received no feedback on their answers. The signal was presented twice from each of the 12 loudspeakers, so that a "run" consisted in 24 randomized trials (24 localization responses). Each subject completed 30 runs. The results showed that (1) individual performance was very stable along the 30 trials, (2) most of the subjects underestimated (rather than overestimated) the azimuth of the source, (3) there was a great inter-individual variability in the size of the absolute error (mean angular distance between actual and estimated angle), (4) there was an effect of handiness on spatial asymmetries observed on the absolute error: While the error was of similar size for left and right sources for 8 out of the 10 left-handed subjects, this error was significantly greater with right sources than with left ones for 16 out of the 22 right-handed subjects. In a second experiment, we showed that the use of a visual-auditory feedback eliminated this asymmetry after a few (less than 10) learning runs. Future experiments will aim to identify the cause of these spatial asymmetries, the best way to eliminate them, and learning methods for localization distorted acoustical environments