Loving people 'for who they are' is a myth; what you actually do is the only thing that matters in love, and the minute you stop doing whatever it is you did that made you loved, you will no longer be so loved, regardless of what was promised.
Numerical identity does play a role, though. Even if you do come to love someone because of what that person does, you usually don't come to love everyone who does the same things. So, though you love someone, not because he or she is the one unique individual that he or she is, but because he or she is the kind of person he or she is, the *object* of your love still is that one individual (though the *reason or cause* for the love is not that individuals uniqueness but his or her be the right kind of person).
Sometimes however, it seems to me, love is not based so much on what the beloved does as on what the loving subject does (or thinks); at least I guess that this is so in many cases of non-erotic love: parents often do not love their children because the children are particularly sweet but because they themselves take such care of them and always have.
And, indeed, if one's love for a particular person is a kind of behaviour, and if that behaviour has been rather erratically reinforced by the person one loves, we should expect it to be quite resistent to both negative reinforcement and extinction, so that you keep loving people long after they ceased to be the kind of people that made you love them. -- I'm afraid that this is not seldom for the worse of the lover.
If love is a behaviour then all thoughts are behaviours, and I'm not sure I would go that far, although expressing love definitely qualifies. But good point that there is a certain inertia involved due to how difficult it can be to undo conditioning, so it can take a while for the consequences of a change in behaviour to play out — sometimes years. Sometimes long, miserable years. I would point out regarding parents and children that their extremely close genetic relationship might cloud the definition of 'self-interest'. Also, it wasn't my intention to argue that people aren't unique. Nor that we can't love whatever collection of behaviours is unique about them.