I wonder if there's an aspect of the experimental design that's left out of the linked article. For example, could the toys have been hidden behind some sort of

I wonder if there's an aspect of the experimental design that's left out of the linked article. For example, could the toys have been hidden behind some sort of shield, so that the baby couldn't see what color they were picking until they got to the shield? Otherwise, there doesn't seem to be any basis for concluding that the initial choice was random.