Implementation of a Semi-Automatized Interval-Timing Analysis in Time-Reference Memory to Assess the Necessity of Fixed-Interval Repetition.

The neurobiological basis of time estimation remains poorly understood. One hypothesis proposes that temporal estimation recruits cognitive processes such as attention and/or working memory. We propose an alternative view: time estimation could constitute a form of immediate/short-term memory in which time itself acts as the stimulus that triggers a behavioral response. To test this, we developed an automated interval-timing paradigm for Drosophila melanogaster in which isolated, head-fixed flies are offered nine sucrose drops, each available for 10 s and accessible only through a Proboscis Extension Response (PER). Behavior is recorded with an overhead camera, and DeepLabCut is used to track proboscis length and drop position over time. Here we present the initial implementation of a semi-automated analysis pipeline aimed at determining whether repeated exposure to a fixed inter-drop interval is sufficient to induce interval timing. From normalized temporal series, we compute dynamic kinematic features, including instantaneous velocity, acceleration, movement direction, and distance to the stimulus, for each PER event. These features will be used to characterize the temporal structure of proboscis extension across trials and to test whether responses come to anticipate drop onset/offset as training progresses, a signature expected if flies learn the fixed interval rather than merely responding to the stimulus itself.