Jennifer Bartlett¹ (), 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. Traditionally, attentional templates represent the exact features of an object; however, sometimes search relies on a template that is not an exact match to the search target but instead predicts target features. For example, when doing a puzzle, the piece you are looking at is not an exact template for the target piece you are looking for. What do we represent in VWM when relying on a predictive template? We evaluated whether participants represent the predictive template in VWM to compare with perceptual inputs or if they generate the target features in VWM to guide search directly. In two experiments, participants first viewed a predictive template (a solid black puzzle piece (E1) with an arrow indicating the position of the connecting piece they were to search for). Most trials then ended in a search for the target (connecting) piece. Critically, 40% of trials instead ended in an unrelated search task that contained a puzzle-piece distractor meant to be ignored. This distractor could either match the predictive template, the target (connecting) piece, or be novel to the trial. This distractor induced memory-driven capture (a slowing of response times when an item matches the contents of VWM) only when it matched the target, suggesting that a representation of the target features was active in VWM in preparation for search. In E2, we replicated this result with puzzle-piece stimuli that had complex designs overlaid on them and thus required pattern extension in order to generate target features. We suggest that predictive templates prompt the generation of target features in VWM to directly guide search, both when the target features are simple and complex.