The application of thought experiments to moral philosophy rather than, say, physics seems oddly pernicious. You need to accept a number stages of abstraction in order to go from lived experience (the first of these abstractions) to the notion of a person in general as seen from the outside - the idea this is the right object of consideration. In trolley type problems, for example, all of this is compressed into a kind of shorthand whereby it is tacitly assumed that the objects of moral concern, are plural and can be enumerated. Both consequentialists and deontologists are assumed to agree that one other person is metaphysically different from ten other persons, and that this distinction is admissible evidence for moral deliberation, even if ultimately discounted. If you were to take a step back and think of the rule of substitutability, then it would make no difference whether you were the one or one of the ten. And stepping further back, why should the details of the relationships already existing among these twelve hypothetical personages be irrelevant? Why should you not favour, or not expect to be favoured by, those closest to you? The question is begged by the set-up of the imaginary scene. The morality of the situation, the implication that an imperative is at stake, only serves to reinforce the authority of the abstraction into a multitude of interchangeable and indistinguishable units - the belief that what you are is what you are represented to be.