Pupillary response indicates the resolution of proactive interference in a visual working memory task

Proactive interference (PI) occurs when previously learned, currently irrelevant information competes with relevant information during retrieval, leading to lower accuracy and/or longer response latencies. Most prior research on PI used verbal stimuli (word lists, e.g., Johansson et al. 2018), but recently, strong PI effects were found in visual memory as well (Endress & Potter, 2014). However, Makovski (2016) demonstrated that memory may be resistant to PI when to-be-remembered items are presented in distinct locations. Here, we use a visual working memory (VWM) paradigm to test (1) whether PI can be induced for spatially-distributed items, and (2) whether cognitive effort (as indexed by pupillometry) reflects PI resolution. Observers (N = 33) performed a VWM task while eye movements and pupil diameter were monitored. In each block of trials, four isoluminant, grayscale objects were presented side-by-side for 2s. PI was induced by repeating two of the four items across trials (the other two were novel). After stimulus presentation, participants completed a brief verbal task to prevent rehearsal. Then, in the retrieval phase, one of the four items (a repeated item during ‘PI trials’, and a novel item during ‘no-PI trials’) was presented again and participants were asked to indicate its prior location by button press. Response times were limited to 2s to avoid a speed-accuracy tradeoff. (1) We found that PI had a significant effect on accuracy [t(32) = 2.62, p < 0.05], indicating that VWM is susceptible to PI even when repeated items are in different spatial locations. (2) We found that participants’ pupil diameter during retrieval was higher in PI trials [t(32) = 3.19, p < 0.01] indicating that, consistent with Johansson et al.'s (2018) results using verbal stimuli, more cognitive effort may be deployed during retrieval when interference is present.