Stereo Camera - Convergence Distance

I got a question concerning the convergence distance setting for stereo camera lenses.
What does this distance actually do? Are all elements between the camera and the converge point in front of the display and the rest inside of it?
I tried playing with the converge during runtime and couldn’t get the scene to pop out of the display.
I had the object at (0,0,0), the camera at (0,-10,0) and the initial convergence distance at 10.

I thought the images of the objects that should be in front of the screen have to be “switched”, so that the left view is to the right side of the right view and vice versa. When I increase the distance the images move inwards but won’t cross each other no matter how high I set it.

The outward movement gets faster with lower distances while the inward one gets slower with higher ones. So it seems to me like the functionality is somehow switched. Or do I have to start the converge at -10?

Yes, generally, things closer than the convergence distance (which should be a positive number) will appear to be in front of the screen. The interocular distance is also relevant, however; if you set this too large, the two views may never appear to cross. This is because the stereo effect is created with parallel frustums rather than toe-in frustums, which generally creates superior stereo, but is limited in how extreme you can make the stereo effects when your interocular distance is not much smaller than your convergence distance (i.e. when you’re putting a fly directly on the bridge of your nose).