This is a really interesting article and includes some studies which might show why girls are put off from certain (STEM- based) games or toys at a very young age. Read more: https://www.theatlantic.com/science/archive/2017/01/six-year-old-girls-already-have-gendered-beliefs-about-intelligence/514340/
A short description was read out to children aged 5 to 7 and the children asked to guess who was being described from a selection of 4 photos (2 male, 2 female). The story was this: “There are lots of people at the place where I work, but there is one person who is really special. This person is really, really smart. This person figures out how to do things quickly and comes up with answers much faster and better than anyone else. This person is really, really smart.”
Among the 5-year-olds, both boys and girls associated brilliance with their own gender. But among those aged 6 or 7, only the boys still held to that view. At an age when girls tend to outperform boys at school, and when children in general show large positive biases towards their own in-groups, the girls became less likely than boys to attribute brilliance to their own gender.
Why do these beliefs occur? It’s not to do with actual ability. At that age, girls tend to outclass their male peers—and the girls in this study knew it. When she showed them pictures of four children and asked them to guess who got the best grades, the older girls were actually more likely to pick girls than the older boys were to pick boys. “Everyone agreed that girls do better in school but that didn’t seem to matter."
Those girls who had most strongly assimilated the stereotype of male brilliance showed the lowest interest in a game for children “who are really, really smart” rather than in games for children who “try really, really hard”. They had already mentally assigned themselves to Hufflepuff instead of Ravenclaw.
Read more: https://www.theatlantic.com/science/archive/2017/01/six-year-old-girls-already-have-gendered-beliefs-about-intelligence/514340/