Some people find it easier to use 'RMI' by harnessing their preexisting skills with compassion, empathy and acknowledgement of shared purpose using the 'selves' metaphor. This is not dissimilar with the way you talk about 'monkeys and horses' and 'giants and tricksters' as a teaching mechanism.

Of course. And in that context, I also teach self-empathy (ala Vladimir's example of dialog in other comments on this post). But I frame that in terms of behavior and metaphor, not stating that there "really are" such entities as giants and tricksters and monkeys and horses. The stories of the Giant and the Trickster, and the Planet of the Horse-Monkeys, were couched in fairy-tale language precisely because they are metaphor, not fact.

But I've mostly found that the flip side to the benefits of these metaphors is that the people who have the most problems are also the ones most likely to abuse these metaphors in a way that keeps them stuck. So, I am really conservative in what I want to say in a context where somebody is asking me (implicitly so here on LessWrong) about what is "true".

Because what is true is that I don't know what goes on in brains. I only know how to describe experience and behavior metaphorically, "as if" the brain had these parts.

I also know from experience that virtually any model you imagine the brain to behave "as if", you can get people to make it come true, by thinking and acting "as if" it were true. This means you want to be exceptionally careful in the models you propose to people you are trying to help, and make sure that you define models that will help them, rather than ones that will keep them stuck.