Abstract
When performing any task for an extended period of time, attention fluctuates between good and bad states. These fluctuations affect perception and performance in the moment, but may also have consequences for what gets encoded into memory. Here we test this hypothesis by monitoring attentional fluctuations in behavior during a response-inhibition task and exploiting them to predict and manipulate subsequent memory. In the response-inhibition task, participants categorized scenes as indoor or outdoor by pressing one of two buttons. Critically, 90% of images were from one subcategory (e.g., indoor) to encourage habitual responding and increase the error rate when the images from the other subcategory (e.g., outdoor) appeared. Sustained attention was defined based on response times (RTs) to the dominant subcategory, as previous studies have shown that errors for the rare subcategory are preceded by relatively fast RTs. Thus, we operationalized a bad attentional state as a period of fast RTs and a good attentional state as a period of slow RTs, relative to the mean RT. Participants then completed a surprise recognition memory test in which they were shown old images of both subcategories from the response-inhibition task and new lure images. In Experiment 1, we found that attentional state during encoding correlated with subsequent memory, with images encoded in a bad state more likely to be forgotten. In Experiment 2, we tracked attentional states in real time and triggered the presentation of an image from the rare subcategory when participants strongly entered a state. Images triggered in a bad state were more likely to be forgotten than images triggered in a good state. These findings suggest that sustained attention has downstream consequences for what we remember. By adapting the experimental design based on participants' behavior, we were able to establish a stronger causal link between sustained attention and memory.
Meeting abstract presented at VSS 2016