Adibi, M.

Motion perception depends on estimating the relative timing of sensory events under internal uncertainty. Although perceptual uncertainty is commonly represented by a single internal noise parameter, its structure, sources, and temporal organisation remain poorly understood. Here, I investigated the computational structure of internal uncertainty in phase-based tactile motion perception using continuous amplitude-modulated vibrations delivered simultaneously to two fingertips. Motion discrimination accuracy, response latency, confidence, and confidence entropy exhibited systematic phase-dependent changes, revealing distinct behavioural signatures of temporal uncertainty. Computational analyses demonstrated that tactile motion perception is not explained by a single source of landmark timing uncertainty. Instead, behavioural performance was best accounted for by two additive uncertainty components: an amplitude-dependent component associated with extracting temporal landmarks from the vibration envelope, and an amplitude-independent component shared across stimulus conditions. This dual uncertainty framework consistently explained the frequency dependence of motion perception, the reduced uncertainty observed for sharper vibration envelopes, and the previously reported enhancement of motion perception with exponential compared with sinusoidal modulation. An independent temporal order judgement experiment further validated the uncertainty parameter inferred from motion discrimination, demonstrating that sharpening temporal landmarks reduced timing uncertainty by 35%. Finally, manipulating the initial stimulus state showed that perceptual choices followed the temporal sequence and correspondence of landmark events rather than the initial evolution of the vibration envelopes, providing independent support for landmark-based computations. these findings demonstrate that tactile motion perception is governed by multiple computational sources of landmark timing uncertainty and establish phase-based tactile motion as a tractable paradigm for independently measuring, manipulating, and modelling the computational structure of perceptual uncertainty underlying sensory decisions.