The important part is how well tuned. In formal terms, how big is the overlap and effect size. If overlap is big or effect size is small, the difference will

The important part is how well tuned. In formal terms, how big is the overlap and effect size. If overlap is big or effect size is small, the difference will be irrelevant for most tasks. The question on thresholds for these values requires further major study. Such as devising a wide set of tasks and comparing performance in real tasks that putatively use it vs these measured motion detection differences. I'd start with catching a thrown ball. Avoiding getting touched by a moving object. Matching speed by movement. Etc. Additionally, it has to be seen if specific training can change those values and to what extent.