Abstract
Dynamic switching of attention between voices in multitalker situations is often investigated in paradigms that combine selective listening with 'attention switching'. Participants are presented concurrently with two talkers, a female and a male, and asked to respond to the number spoken by the talker specified on each trial by a cue. A change in the target voice (when compared to listening to the same voice) results in a robust performance 'switch cost'- which can be reduced substantially by increasing the preparation (cue-stimulus) interval. Using dichotic presentation we asked whether preparation also increases the selectivity for the cued (relevant) voice dimension- gender (in one session) or location (in another session). We examined the interaction between the features of the relevant dimension and features of the irrelevant dimension (which varied independently) as a function of preparation. When the two voices (genders) were heard from the same locations as on the preceding trial, performance was better than when genders swapped locations relative to the previous trial- suggesting 'binding' between genders and locations. The key question was whether preparation reduced this binding effect- which would indicate greater dimensional selectivity. We found the opposite- the binding effect was significantly larger when there was more time for preparation. Since preparation reduced the switch cost but increased the binding effect, the results reveal a dissociation between the effect of preparation on the switch cost and on the binding effect. We propose mechanisms by which preparation may enhance the formation of bindings and/or their retrieval.