Abstract 1294, Date 2:00 PM - 4:45 PM, Wednesday, February 25, 2004 Session T:Psychophysics: Learning and Temporal Processing
Mechanisms of Perceptual Learning in Amplitude Modulation Detection
Ying-Yee Kong, Zhong-Lin Lu, Barbara Dosher, Fan-Gang Zeng
Unlike cognitive learning, perceptual learning is stimulus and task specific. We adapted the theoretical framework developed by Dosher and Lu (1998; 1999) in visual perceptual learning to identify mechanisms of perceptual learning in auditory amplitude modulation detection. By measuring an observer's performance in the presence of external noise at systematically manipulated levels, the perceptual template model characterizes the observer's perception in terms of equivalent internal noise, identification templates, and processing nonlinearities. Mechanisms of perceptual learning can be characterized as improvements of these system limitations, including stimulus enhancement, external noise exclusion, and/or internal (multiplicative) noise reduction. Four normal-hearing listeners were tested. The signal was an 8-Hz modulation of a 0-dB SPL sinusoidal carrier (6.5k Hz). All subjects were trained in 10 or more sessions in the modulation detection task tested in quiet and five noise spectrum levels (-10, 0, 10, 20, and 25 dB SPL). Modulation detection thresholds were obtained using two three-interval forced-choice procedures (2-down 1-up and 3-down 1-up) in each session. Each session included 60 trials per 2/1 staircase and 80 trials per 3/1 staircase for each of the 6 noise levels, yielding 840 trials per session. In addition, detection thresholds of the unmodulated carrier in the six levels of external noise were measured before and after perceptual training. Results showed that (1) training did not improve carrier detection, (2) training reduced modulation detection thresholds in quiet and near quiet conditions by about 3. dB, and (3) training did not change modulation detection thresholds in the high noise conditions. These findings imply that modulation detection training enhanced the stimulus (or, equivalently, reduced the internal additive noise) but did not optimize the identification templates, nor change the nonlinear processing in the system.       Supported by NIH RO1-DC0226