Abstract During conversations in the presence of other competing talkers, multiple speech streams compete for listeners’ attentional focus. Listeners must segregate speech streams, and selectively attend to the target talker while filtering out irrelevant speech. Research on value-driven attention suggests that perceptual and attentional processes are biased by prior rewarding experiences with stimuli—that is, high-valued stimuli win attentional competition. However, it remains unknown whether such effects generalize to speech perception in multitalker environments. Here, the present study investigated whether listeners can better understand speech spoken by talkers associated with higher versus lower values in challenging listening conditions. In three experiments, we used reward-based training paradigms to either explicitly or implicitly induce listeners to associate talkers’ voices with varying magnitudes of rewards. Subsequently, listeners performed a speech-on-speech intelligibility task in which the exposed voices were either the target or distractor. We found that explicit talker–reward learning did not influence speech intelligibility. In contrast, we observed that implicitly acquired talker–reward associations improved the intelligibility of high-reward talkers, but only in the most adverse listening condition. However, this effect did not replicate when we further imposed greater attentional demands by presenting target and masker talkers’ speech without spatial separation. Instead, listeners relied primarily on acoustic cues other than reward-associated talkers’ voices. Furthermore, the reward values associated with the distractor talker had no effect on interference. These findings suggest that value associated with talkers’ voices do not reliably bias listeners’ attention to overcome the complex, acoustic constraints imposed in speech-on-speech perception.
Lim et al. (Fri,) studied this question.