The case I had in mind involved enabling a drug addict to get more drugs. The only gain was that the addict became less unhappy for a time. Anyway, I propose t

The case I had in mind involved enabling a drug addict to get more drugs. The only gain was that the addict became less unhappy for a time. Anyway, I propose that given that humans (possibly barring autistics etc) operate by building models of the minds of other people inside their own head, if a person does something based on that model to help the modelled person become more happy, it's a selfless act - I can quite literally create a model of someone else inside my head, calculate how to make them happy based on that model, and then do it. I'm not modelling how I can make myself feel good, or modelling how I'll look like a big man for helping someone; I'm modelling another person's emotions to work out how to make them happy.