Models of attention posit the existence of a target, or attentional, template in working memory. Traditionally, the template has been characterised as containing a static and veridical representation of the target that is most efficient when highly precise.

In this talk, I will share data from my lab that challenges this notion and instead suggests that the target template is highly flexible and contains information that maximises the ability to distinguish targets from distractors. This information is often off-veridical and determined by expectations about the local stimulus context, target variability, and learned associations. Our results indicate that the purpose of the target template is to maximise target-to-distractor distinctiveness and doing so often relies on information beyond the visual features of the target object.