Abstract
<p>Mouse tracking is used to infer how competing response tendencies evolve from initial activation through online action control. Such psychological inferences depend on the trajectory representation chosen before analysis. We distinguish two reference paths, the line joining each trajectory's observed endpoints and the fixed path defined by the experimental layout, and three alignment schemes. Algebraically, own-endpoint deviation is zero at both endpoints. Start alignment subtracts each trial's initial offset, and start-plus-end alignment fixes both endpoints. Time-course contrasts after these operations cannot identify effects at the constrained points. In simulations calibrated to an empirical dataset, all six pipelines recovered effects that arose and resolved inside the movement. When the true effect was largest at movement onset or at the response, own-endpoint pipelines displaced the apparent peak to the opposite end, while start alignment converted an onset difference into a late difference of opposite sign. An unconstrained design path recovered participant-level onset and terminal components at r = .94 and .95. Varying resampling, smoothing, noise autocorrelation, endpoint variability, and alignment geometry preserved these conclusions. In a Simon-task illustration (38 participants, 4,684 trials), incongruency delayed initiation by 10.38 ms and movement by 41.59 ms, consistent with conflict extending into action execution. Yet the same trajectories supported different accounts of early activation and late correction after alignment. Spatial preprocessing therefore defines which psychological process claims remain testable. We provide an estimand-based decision guide, a robust movement-onset diagnostic, and reproducible code.</p>