Tomcat Visual representation simplifies perception.
Coding is not about perception. And you've yet to prove that the process of coding gets simplified in any way by using boxes and noodles.
The idea that "visual coding" simplifies coding is based on a fallacious, naive analogy that, duh, picture books are "simpler" than regular books ergo less text makes things universally "simpler". The problem with that analogy is that - coding is nothing like reading a book. Replacing text with "pictures" does not simplify it. Neither does replacing typing with clicking because programming is not about entering data into a computer either.
Besides, any box coding system still uses quite a lot of text. It's not really "visual representation". It's a form of textual representation with 2D layout control. Ironically, in many cases the amount of total text displayed by a box system is much higher than the amount of text needed to do the same thing in proper code.
As an example, try to step a component of a vec3 in Godot's visual shader and compare how much text you need to do the same with code. Here, let me do it for you:

vs.
step(0.5, myvector.x)
Do I need to count the characters as well? Oh, ok. The "visual representation" uses 500% more text.
On top of that, when you go beyond typical trivial examples, the node network can become an enormous visual mess and you need to spend additional mental effort on spatial management of your boxes. This can become significant as the complexity of the system increases. Not to mention how much screen estate is wasted on non-functional empty space inside those boxes.
You're insinuating that "visual coding" is somehow a prerequisite to learn proper coding. This is again fallacious. There is no such prerequisite. Millions of people have learned to program without ever seeing any "visual code".