Xiaonan Li¹ (), Moxuan Liu¹, Yaxin Liu², Stella F. Lourenco¹; ¹Emory University, ²Georgetown University

The (cognitive) pupillary light response has emerged as an index of visual imagery. When tasked with imaging dark and light scenarios (e.g., night vs. day sky), people’s pupils dilate and constrict, respectively. As an assessment of visual imagery, pupillometry is advantageous because it bypasses the need for conscious reporting. Yet, even with pupillometry, explicit instructions or cued words are generally included to elicit visual imagery, which may confound imagery-related processes with other (non-imagery) processes such as metacognitive awareness. Here we develop a novel, implicit pupillometry-based paradigm, in which there are no instructions related to visual imagery. This paradigm leverages predictive learning during the occlusion of a moving object. In an initial familiarization phase, participants repetitively viewed a coin-shaped object that moved horizontally across the screen (e.g., left-to-right), while rotating between its front (bright) and back (dark) sides. In a subsequent test phase, the object rotated behind an occluder toward the end of its trajectory. On half the trials, the object rotated from dark to bright and, on the other half, it rotated from bright to dark, with luminance during occlusion matched across trials. Analyses of perception and imagery periods with 9 participants revealed robust pupillary light responses to dark and light stimuli. When tracking visible dark and light stimuli, pupil sizes exhibited dilation and constriction, respectively. Moreover, during the occlusion phase, baseline-corrected pupil sizes were significantly larger for the predicted dark stimulus than the predicted bright stimulus, suggesting that participants engaged in visual imagery consistent with the learned luminance associations. These findings support the validity of a novel implicit task of visual imagery in which imagery occurs without explicit instructions or conscious reporting. This task has the potential to provide a powerful tool for investigating visual imagery across diverse population, significantly advancing the study of visual imagery.