Reward Maximization Is Limitation With the Homeostatic R L Framework

Hawdos: Would there be any r l model with an agent's internal set point or internal state dependent on context from the environment? And along those lines, how could the multi dimensionality of value map, or enternal drive, like thirst versus hunger, be computed in the a i agent o. I would love to hear also from farmat and from thers what they think about these competing dimensions of reward that i think to some extent ashot mentioned. Think actually, there are different dopomi for different dimensions of reward.