"I've never understood Maryam," she said. My heart dropped. I turned pale, felt helpless, and left hopeless. It was a big, pernicious lie that I had kept telling myself, there has never been any transitive relation between we three.
I guess understanding happens at a deeper level... I mean at the very least I think you need both willingness of individuals to reveal themselves and time to do so... It would be interesting though to study the types and conditions under which something like understanding is transitive (I'm thinking along the lines of for example trust in belief being transitive and trust in performance not being transitive)
I remember once Peter Norvig, the coauthor of our AI book, said, in one of his online lectures, that if you don't understand somebody, you should assume that this agent is anyway a rational person, perhaps his goal or utility function that he uses is different from yours. Inspired by his AI lesson, I think the person who does not understand Maryam does not understand her goal or her values. I'd rather infer this person has never had the same or similar goals and values as Maryam has. So I suppose if B and C feel that they share some thoughts or values, but A and C apparently do not share the same values, I'd infer that the relation between A and B has not ever been based on those values. And it is depressing, for B! ;)
I don't know Sahar... I would think if one has enough *time*, one can to a good approximation *learn* someone else's set of preferences and utility function (assuming people act based on those) even if they are not similar to hers. I also think understanding someone's acts is different from understanding someone's goal or utility function, since in the first case all you need is to know about their goal or utility function, but in the second case you need something at a higher level.
I guess understanding happens at a deeper level... I mean at the very least I think you need both willingness of individuals to reveal themselves and time to do so... It would be interesting though to study the types and conditions under which something like understanding is transitive (I'm thinking along the lines of for example trust in belief being transitive and trust in performance not being transitive)
ReplyDeleteI remember once Peter Norvig, the coauthor of our AI book, said, in one of his online lectures, that if you don't understand somebody, you should assume that this agent is anyway a rational person, perhaps his goal or utility function that he uses is different from yours. Inspired by his AI lesson, I think the person who does not understand Maryam does not understand her goal or her values. I'd rather infer this person has never had the same or similar goals and values as Maryam has. So I suppose if B and C feel that they share some thoughts or values, but A and C apparently do not share the same values, I'd infer that the relation between A and B has not ever been based on those values. And it is depressing, for B! ;)
DeleteI don't know Sahar... I would think if one has enough *time*, one can to a good approximation *learn* someone else's set of preferences and utility function (assuming people act based on those) even if they are not similar to hers. I also think understanding someone's acts is different from understanding someone's goal or utility function, since in the first case all you need is to know about their goal or utility function, but in the second case you need something at a higher level.
ReplyDelete