Individuality is an intrinsic and essential aspect of mammalian behavior that emerges even in genetically identical organisms experiencing the same environmental conditions1. In the International Brain Laboratory, over one hundred mice were trained on a visual decision-making task with the explicit goal of establishing a rigorously standardized experimental protocol. This effort led to an automated pipeline that produced trained mice whose behavior was indistinguishable across seven different labs, when considering trial-level descriptors of behavior. Nevertheless, substantial inter-individual variability was evident in both training time and proficient behavior, but its nature remains poorly characterized. To address this, we developed a semi-supervised behavioral segmentation approach to characterize mouse behavior with 100 ms resolution across multiple variables, including wheel movement velocity, whisker motion energy and licking. This yielded a discrete space of behavioral syllables which we further analyzed in the context of the trial structure. Variability in the expression of behavioral syllables was highly non-random, revealing structure in how different behavioral features co-vary at the sub-trial level. Moreover there were further indications that mice fell into several clusters, indicative of strategy types or even mouse personality types. The micro-behavioral structure derived from trained mice was further informative of differences in learning speed across individual mice, supporting its stability and biological significance. Overall, these results provide evidence that even in a cohort of mice whose overt task performance behavior is indistinguishable, there exist latent variables, manifesting in the details of micro-behavioral features, which appear to explain important aspects of behavioral individuality