A related paper here. Explains 7+ layers of nesting is actually possible to keep track of, but I can show 20 layers is not out of the question in some situations. https://www.researchgate.net/publication/271539325_The_ease_and_extent_of_recursive_mindreading_across_implicit_and_explicit_tasks

Another common but not often mentioned approach to thought is provisional reasoning about new paradigms, where you provisionally accept a claim as 100% true and then look at how well the known facts interlock in that scenario, seeing if it altogether ends up making more sense than the existing belief system /paradigm (rather than questioning each piece in turn with reference to the current paradigm, which heavily biases one in favor of it).

An example would be imagining your stock portfolio has fallen greatly a month from now and seeing how much sense the currently known facts would make in light of that, if in that scenario you would say the writing was on the wall or not; then imagining it has risen greatly a month from now and doing the same; then imagining it has stayed roughly the same; then seeing which one altogether makes more sense. Compare the plausibility of the two at the end, as unified wholes, rather than midway through. Very few seem able to avoid the error of switching scenarios midway through.