Retrieval-related eye movements are predictive of memory precision.

Eye movements are predictive of successful episodic memory encoding and retrieval, but it is unclear whether they reflect the precision of retrieved memory content. Here, we examined relationships between eye fixations, memory precision and fMRI BOLD activity. At study, participants were presented with object images, each placed at a random location on an invisible circle. At test, both studied and new images were presented. Participants were instructed to make a covert recognition memory judgment to each image, recall and then signal its studied location, guessing if necessary. Simultaneous fMRI and eye-tracking data were acquired during the test phase. For correctly recognized images, the trial-wise angular distance between the studied and reported locations was fitted to a two-component mixture model. Based on the model-derived parameters, correctly recognized images were categorized as those associated with successful location retrieval (location hits) or guesses. Location hits were further divided into high- and low precision trials. At retrieval, fixations predicted both successful retrieval and the precision of the retrieved location memory. Additionally, across-trial fixation patterns were more similar for location hits than for guesses, and for high- than for low-precision trials. The association between retrieval success and eye fixation precision was evident across almost the entire recall phase, whereas memory precision effects emerged around 1-1.5s after image onset. Fixation precision and fixation pattern similarity each predicted trial-wise memory precision independently of hippocampal BOLD activity, which also predicted precision. These findings suggest that eye fixations during retrieval track the fidelity of mnemonic content.