Jennifer Bartlett¹ (jdbartlett@mun.ca), Blaire Dube¹; ¹Memorial University of Newfoundland

During visual search, a representation of the search target (an attentional template) held in visual working memory (VWM) supports the automatic guidance of attention to feature-matching perceptual inputs. Previous work has demonstrated that the precision of these attentional templates - how closely the representation held in VWM matches the features of the search target - affects the efficiency of search performance: As templates become less precise, observers adopt broader attentional sets that guide attention less accurately. In the present study, we evaluated whether we could narrow the attentional set and mitigate the response time (RT) cost associated with using imprecise templates by providing predictive cues that would describe how the search target would differ from the initial template image. Participants searched for real-world objects/animal stimuli among an array of distractors under three cue conditions: precise template cue, where the target matched the initial template exactly (“search for this exactly”); imprecise template cue, where the target features were similar to the template, but with a change in state that was unknown to the observer (e.g., a template depicting a bird perched but a target stimulus that was a bird in flight; “search for something similar to this”); and predictive template cue, in which the target features were similar to the template, but the change in state from template to target was predictable based on a provided cue (“search for this in flight”). Predictive cues significantly reduced the RT cost seen when using imprecise templates, but RTs in the predictive template condition were still significantly higher than in the precise template condition. These findings suggest that predictive cues can narrow the attentional set somewhat, partially recovering the cost associated with using imprecise attentional templates and allowing for more efficient search; however, templates generated with predictive cues are less efficient than precise templates.