Notwithstanding the neuromyth controversy, the malleability of learning style preferences impacts the validity of the measurement instrument and the effectiveness of the associated model of learning. This study investigates the test-retest reliability and underlying dynamics of Kolb's Learning Style Inventory (KLSI). It surveys 245 college-level students in Australia, over three rounds of data collection at 7-week intervals. Results show that over 75% of participants are incorrectly categorised, and more than 50% materially changed their category of learning style between rounds. The study reveals that individuals roam freely, rapidly, and extensively across learning style categories. Thus, the categorical differences measured by the KLSI lack meaningful purpose. Whether or not learning styles are a neuromyth, this study indicates that the act of learning, as an act of agency, is fluid with potentiality and choice. The more meaningful focus for teaching and learning practice would then be on student commonalities, not categorical difference.