Lily Hallenbeck¹, Jamal Williams²; ¹UCSC

Perceiving and utilizing temporal intervals is critical for everyday life. Crucially, these abilities rely on multiple senses working together. Despite the multimodal nature of real-world time perception, we have a limited understanding of how we maintain and transfer time between modalities. This project addresses a fundamental and unresolved question: Can a duration learned through one sense (e.g., audition) be used to guide behavior in another sense (e.g., vision)? To answer this, participants encode a short, target-duration in working memory for a later test. Target-durations are filled intervals that are presented as either a colored rectangle on a black background (visual), a continuous, complex tone (auditory), or both (audiovisual). While maintaining this duration, participants detect small bursts of coherent motion in a visual field of randomly moving dots. Importantly, the timing of these coherent motion bursts occasionally coincides with the interval in memory: dots jitter for the target-duration, followed immediately by coherent motion. In this task, performance is worse when motion bursts do not align with visually encoded target-durations, but not via auditory nor audiovisual intervals. These results suggest that temporal intervals guide behavior within, but perhaps not across, modalities. To further compare the quality of temporal-durations, the recognition memory task can be presented in the original modality (e.g., audition-audition) or across-modalities (e.g., audition-vision). Within modality, we find that memory performance is best for auditory intervals, worst for visual, and suffers only limited costs when transferring from vision to audition. In contrast, when an auditory interval is tested using a visual probe, performance is near chance. In tandem, these findings suggest that auditory memories are best remembered but that visual intervals are best for guiding behavior. These findings help inform competing models of temporal perception, including central clock theories and distributed, modality-specific timing models.