Fixation sequences in imagery and in recognition during the processing of pictures of real-world scenes

Abstract

Thirty photographs of real-world scenes were presented for encoding, and half the participants then performed a recognition test, deciding whether each of 60 images were old (from the original set) or new. The other participants performed an imagery task immediately after encoding each of the 30 images. After completing this task, the recognition group then performed the imagery task in response to prompts that were unique verbal descriptors, and the imagery group performed the recognition task. All participants returned 2 days later, and repeated the imagery test. Eye movements were recorded during all phases. Differences in average fixation duration, average number of fixations and average saccadic amplitude were found between task groups and between experimental phases. Scan patterns were compared with a string-editing algorithm. Close similarities were observed between experimental phases that involved more similar tasks (e.g., initial encoding vs. recognition, and immediate imagery vs. delayed imagery). Scan patterns were equally similar when the task was presented immediately or after 2 days. We propose that the more similar the encoding and retrieval processes are, the more similar eye movements will be at each of these experimental stages.