3) This indicates that imagery can modify memory to accommodate anticipated changes, improving the ability to detect that a familiar face is present—but not the ability to pick that face out of a lineup. These findings thus identify a novel dissociation between old/new and forced-choice recognition.

2) At study, participants saw neutral faces and were cued to imagine them in happy or angry expressions. At test, old and new faces were shown as happy or angry. When old faces' test expression matched the imagined expression, old/new recognition was better—but forced-choice accuracy was unaffected.

1) Given that items don't look exactly the same at encoding and retrieval in real-world recognition—including consequential uses of memory, like eyewitness memory—Darya Zabelina and I examined whether visual imagery could be used to improve our ability to recognize people across appearance changes.

4) These results led us to propose a new theory of attentional guidance, which we term rational integration: different sources of information, in this case episodic memory and semantic knowledge, are rationally combined and prioritized based on their relative strength/precision to guide attention.

3) When only unconscious memory was available—i.e., cases in which participants exhibited memory-driven performance improvements despite a confident lack of awareness for that memory—memory only guided search when semantic knowledge had failed to get the eyes to the target (aka, incongruent scenes).

2) We manipulated semantic knowledge via schema congruency (objects in congruent vs incongruent scene locations), and measured recognition memory for the scenes. When detailed recollection was available, memory was integrated with semantic knowledge to guide early eye movements during search.

1) The extent to which episodic memory guides visual search when semantic knowledge is available is debated. We found that whether memory influences search depends on what type of memory is available.