Charles S. Watson
Xiaofeng Li
Gary R. Kidd
Yijian Zheng
Dept. of Speech and Hear. Sci., Indiana Univ., Bloomington, IN 47405
Training to attend selectively to certain spectral-temporal components of four tonal patterns has been found to have only slight effects on discrimination performance [Port et al., J. Acoust. Soc. Am. 93, 2315(A) (1993)]. In a new version of this experiment, a novel pattern (ten 50-ms tones, 300 3 kHz in frequency) was presented on each trial. One group of listeners was trained to discriminate changes in the early low-frequency region of the patterns, a second group in the late high-frequency regions, and a control group was trained with changes occurring throughout the patterns. Training was conducted for ten sessions, followed by testing at all spectral-temporal positions. Effects of selective training were much more substantial than in the earlier four-pattern experiment. Under the high stimulus-uncertainty conditions, attentional training yields the predicted results: Discrimination is relatively improved for trained, compared to untrained regions. An unexpected result was that the control group's performance was significantly more accurate than that of earlier selective-attention group. It is possible that efforts to selectively attend to some spectral-temporal region of an unfamiliar pattern may reduce overall discrimination performance, compared to that achieved when listening for any change in the pattern. [Work supported by AFOSR.]