Listeners routinely perceive phonetic speech signals which are made up of acoustic detail belonging to multiple continuous physical dimensions (e.g. intensity, frequency, duration), and then discretely map them onto phonological units like phonemes with ease. Traditional accounts of speech perception suggest that listeners achieve this by discarding all non-distinctive (within-category) variability in the signal in favor of discrete phonological representations, resulting in a phenomenon known as categorical perception. However, more recent findings show that listeners do exhibit sensitivity to intra-categorical phonetic detail, for example by investigating on-line measurements from eye-tracking. It is yet unknown whether mouse tracking, a nascent experimental method capable of producing continuous multi-dimensional measurements from motor behavior, can similarly contribute meaningful evidence to research into categorical perception of speech sounds. The present exploratory studies indicate that the effectiveness of the mouse tracking paradigm may be severely limited for such purposes. Here, mouse tracking was unable to replicate previous findings on listener sensitivity to sub-phonemic variability, although a decidedly non-antagonistic replication attempt was made with regards to the design specifics of the paradigm. These findings undermine assumptions about the mapping between cognitive processes and manual response dynamics, questioning the utility of mouse tracking for speech perception research.