Is there any empirical data regarding the usability of websites that (for lack of a better word) "hijack" the (vertical) mouse scroll wheel to perform a task other than scroll up/down the page?
My specific case involves a carousel. It has been requested that when a user moves their mouse wheel up/down while their pointer is over the carousel, the result is that the carousel advances rather than scroll the page up/down. The effect in question is at least fairly popular as I was able to find a few examples (Example 1, Example 2).
My hunch is that this would be frustrating from a user experience because it's unexpected. But I can't find any definitive resource that has proven that. Perhaps it's common enough that it isn't unexpected anymore?
Is anyone aware of an user test that proved or disproved this? Or are there any industry leaders who can at least back up my primarily opinion-based hunch?