But also: it is surprising how well building up the intuition in 3 dimensions works for N dimensions. You can build up a very working, visual intuition for Linear Algebra just by doing everything in 3D, for example. There's a textbook that takes this approach: "Practical Linear Algebra".
But, of course, this comes with a caveat: it works until it doesn't. The volume of a unit sphere increases for dimensions 1-5, then starts to go down[1]. There exists an exotic R^4: a manifold that is homeomorphic, but not diffeomorphic to R^4[2]. And so on.
But for a lot of (intro) differential geometry and topology, (e.g. concepts like covering spaces), I haven't found many examples where intuition based on 3 dimensions breaks.
A friend of mine[3] once told me that geometry is the art of getting correct results from incorrect diagrams. I think that the quote we're discussing is in a similar spirit.
[1]https://en.wikipedia.org/wiki/Volume_of_an_n-ball
[2]https://en.wikipedia.org/wiki/Exotic_R4
[3]Dmytro Karabash
Can you elaborate on this? Do you mean that the point is 'to free up your intuition, abandon your hope of truly visualising this stuff, and make do with a hybrid of simplified visualisation and abstract thought'?