| Many theories of visual attention propose that working memory and perception interact during the processing of complex scenes. In addition, previous research with monkeys has suggested that target objects are preferentially processed during visual search due to template representations in working memory that bias perceptual mechanisms in a top-down manner. This dissertation examines four specific theoretical proposals regarding the role of visual working memory in efficiently processing complex multiple item scenes. First, are target templates stored in visual working memory during visual search? Second, if there are more possible search targets, are more templates stored in visual working memory? Third, does the human visual system also automatically allocate attention to objects that match the contents of visual working memory? And fourth, are targets encoded into visual working memory during visual search? I addressed each of these questions in a series of experiments using a dual-task paradigm. Participants were required to store an object representation in visual working memory and then perform a visual search task while maintaining this representation. The findings of the experiments indicate that many of the extant theories of visual attention oversimplify how visual working memory and attention interact to enable efficient visual search performance. Specifically, it appears that holding a representation in visual working memory is not sufficient to bias attention mechanisms to automatically select similar items as has frequently been proposed. In addition, it appears that either working memory or long-term memory representations can be used to guide attention in a top-down manner during visual search. |
