There are experiments where they raise cats in an environment with only horizontal stripes or only vertical stripes. Cats raised with only horizontal stripes will avoid the rungs on a chair, but run into the legs. Cats raised with only vertical stripes will avoid the legs, but smack into the rungs with a mystified look on their face.

That old person already developed their spatial skills in an actual 3D environment. There is evidence that babies and toddlers need substantial experience to learn to see their environment properly.

In experiments, an infant will accept that an object transformed into a random other object while briefly out of their sight. They will not accept that one object turned into two objects.

This makes sense. A plate is flat from one angle of view and round from another. You change the angle of view, the shape radically changes. But it remains a single object.

Babies don't yet know which transformations make sense, so they will accept that a truck transformed into a clown toy. They will not accept that a truck transformed into two clown toys.

So I think you have that backwards. People with an already developed understanding of 3D spatial relationships based on lots of living in an actual 3D world can relate meaningfully to a virtual environment. This in no way supports the idea that simply seeing a representation on a screen of a 3D environment can help a developing child properly code up their spatial reasoning.