Abstract
Rapid resumption refers to unusually fast visual search that occurs when a search display has been pre-exposed. To measure this, observers are presented with repeated search displays, each 100 ms (“look” episode) that are interrupted by blank displays, each 900 ms (“blank” episode). Look and blank displays are presented successively until the observer responds. Search times of less than 400 ms (rapid resumption) occur with some frequency following the second or subsequent looks but never after the first look. This ability to rapidly resume a search may reflect use of a perceptual prediction mechanism. Here, we asked observers' perceptual predictions would be enhanced if targets had been previously associated with rewards or losses. Previous research has shown that learned reward and loss value of stimuli can yield a selection advantage even in tasks where these contingencies are contextually absent. To test this, we combined an instrumental learning task with an interrupted visual search task. In Phase I, participants learned to associate faces with gain, loss, and no outcome. In phase II, learned faces were presented as targets in multi-item arrays. During the task, all stimuli had equal task relevance but differed in their expected value. In the search task, participants searched for a face defined by gender and reported whether a line of dots (present on each face) was on the left or right. The results indicate that compared to search for familiar targets with no expected value, search for targets previously associated with rewards or losses was faster. However, the benefits of expected value did not affect the frequency of rapid resumption responses. This dissociation suggests that value codes for stimuli do not affect the mechanisms underlying rapid resumption.
ESRC support for J. Gomez-Cuerva