D’-tecting the Beat: Refining d’ in Heartbeat Detection Tasks

Cardiac interoception—the ability to perceive and interpret one’s own heartbeat—is fundamental to emotion regulation, cognition, and allostasis, and is increasingly considered a potential transdiagnostic biomarker. Interoceptive accuracy is typically assessed with behavioral sensitivity indices, such as d’ and mean distance, derived from heartbeat detection tasks (HBT). Although these tasks capture rich temporal dynamics, current indices overlook much of this information. As a result, existing analytical approaches remain unrefined and insufficient to capture the complexity of interoceptive processing, limiting both theoretical and clinical insights. We refined d’ for HBT by adjusting how signal detection theory (SDT) outcomes are extracted and by introducing a pooled z-score approach to reduce reliance on questionable statistical assumptions. Thirty-three healthy participants performed an HBT with two exteroceptive blocks (tapping to a recorded heartbeat) and two interoceptive blocks (tapping to their own heartbeat), while electrocardiograms recorded ground-truth heartbeats. We calculated d’ using three methods: the original temporal-window approach, a window-free design, and a window-free pooled z-score design. Mean distance, a response frequency-based measure, was also computed for comparison. The pooled z-score d’ provided a more balanced influence of SDT outcomes, revealed sensitivity differences across approaches, and showed stronger correlations with mean distance. This framework offers a more robust, assumption-light method for assessing interoceptive accuracy. By enhancing the precision of accuracy metrics, the refined d’ supports more sensitive detection of sensory processing differences in clinical populations and strengthens methodological rigor. Furthermore, combining d’ and mean distance may yield novel metrics that capitalize on their shared sensitivity basis.