Abstract
Temporal binding research proposes that voluntary actions compress perceived time between actions and their outcomes, reflecting an implicit sense of agency. However, it remains unclear whether such effects are stable across action--outcome interval durations and whether they depend on the degree of decisional freedom, such as the availability of choice options. Here, we investigated whether action generation (action vs. no action) and choice availability (one vs. two options) influence interval estimation across multiple action-outcome intervals using a yoked action-baseline design. Across two experiments, action-baseline differences consistent with temporal binding were observed only at the shortest interval (300 ms), where action trials produced relatively shorter interval estimates than matched baseline trials. However, these differences did not generalise to longer intervals and reversed in direction at longer intervals in one experiment. Across the full interval range, baseline estimates shifted more steeply with interval duration than action estimates, producing a crossover pattern in which shorter intervals were overestimated and longer intervals underestimated across conditions. We did not find evidence that choice availability, operationalised as one versus two options, modulated interval estimates in either experiment. These findings suggest that action-related differences in interval estimation are temporally constrained and may arise from an interaction between action-related processes and general perceptual biases in temporal estimation, including regression toward the mean. More broadly, the results show that apparent temporal binding effects in interval-estimation tasks can depend strongly on where tested intervals fall within the broader interval range, highlighting the importance of accounting for interval-dependent estimation biases when interpreting temporal binding effects in interval estimation tasks as proxy measure of the sense of agency.