[ Lay Language Paper Index | Press Room ]
Beverly A. Wright b-wright@northwestern.edu
Northwestern University
Evanston, IL 60208
Popular Version of Paper, 2pAB6, “Training-induced improvements on interaural level difference (ILD) and interaural time difference (ITD) discrimination in human adults”
Presented 2:45 p.m. Wednesday Afternoon, Nov 28
154th ASA Meeting, New Orleans, LA
If you can tell that the music you are hearing is coming from a radio located to your right, rather than to your left, you are taking advantage of your brain’s capacity to recognize that the sound reaching your right ear actually differs from that reaching your left ear in two key respects. One difference arises because your two ears are separated by a distance (the diameter of your head), so it takes the sound a longer time to reach the ear that is farther from the sound source than the ear that is closer to the source. The other difference arises because there is a largish object between your ears (namely, your head), that serves to reduce the sound level at the ear farther from the sound source compared to the ear that is closer to the source. The larger the disparity in the arrival time or level between the ears, the more displaced the sound source is toward the closer ear. Thus, unlike for vision and the sense of touch in which the perceived location of a stimulus is determined by the position it stimulates on the sensory surface (the retina, or skin), for hearing, sound-source location is determined by neural computations of the differences in the arrival time and level of the stimulus as received by your two ears. Here we report data showing that human adults can improve their ability to detect small changes in these time- and level-difference cues with practice, but that the patterns of learning differ markedly between the two cue types.
We (Beverly Wright, Matthew Fitzgerald, and Yuxuan Zhang) gave two different groups of listeners practice discriminating sounds based either on the time-difference, or the level-difference, cue. Both groups practiced ~1 hour per day for 10 days. During each training session, we made 12 measurements of the smallest difference that each listener could discriminate. We then compared the average performance of these trained listeners before and after the training sessions to that of listeners who participated in the pre- and post-training tests, but received no intervening training.
The influence of the training differed markedly between the two trained groups. Both groups showed improvements attributable to the pre-training test. However, the listeners who practiced the level-difference discrimination continued to improve with additional practice, while those who practiced the time-difference discrimination did not. In addition, practice on level-difference discrimination did not subsequently benefit performance on time-difference discrimination, or vice versa. One interpretation of these results is that the training improved the computation of time and level differences at a stage at which these computations are made by separate neural circuitry. If the same circuitry were involved in both cases, then the learning patterns might have been expected to be similar for the two cue types, and learning on one cue to aid performance on the other.
It also seems that the improvements observed for the two cues arose from different types of performance change. When we examined the listeners’ best and worst estimates of performance for each training session, rather than analyzing only their average performance (as is the norm), the patterns of results again differed for the two trained groups. The listeners who practiced the level-difference discrimination showed improvements in both their best and worst daily estimates. The improvement in their best performance indicates that these listeners were able to distinguish sounds at the end of training that they could not distinguish at the beginning. This result suggests that the training led to a fundamental improvement in the brain’s ability to detect these small level differences. In contrast, the listeners who practiced the time-difference discrimination showed no change in their best, but a clear improvement in their worst, performance. Even at the beginning of their multiple days of training, these listeners could detect quite small time differences, but not consistently. This pattern suggests that the training did not change the fundamental ability of the brain to detect small time-differences, but rather improved the capacity to access this fundamental ability reliably. Thus, examinations of how training affects average performance, as well as the best and worst performance on each day, provides insights into the characteristics of the neural circuitry modified during training and the types of modifications that occurred.
Overall, these results add to the evidence that sound-localization ability is not rigid, but rather can be modified with experience, raising the possibility that perceptual training could be used to improve this skill in both clinical and non-clinical populations. [Supported by NIH/NIDCD.]