Why wouldn’t you expect terminal values to charge? Does your agent have some motivation (which leads it to choose to change) other than its terminal values. Or is it choosing to change its terminal values in pursuit of those values? Or are the terminal value changing involuntarily?
In the first case, the things doing the changing are not the real terminal values.
In the second case, that doesn’t seem to make sense.
In the third case, what we’re discussing is no longer a perfect rational agent.
Why wouldn’t you expect terminal values to charge? Does your agent have some motivation (which leads it to choose to change) other than its terminal values. Or is it choosing to change its terminal values in pursuit of those values? Or are the terminal value changing involuntarily?
In the first case, the things doing the changing are not the real terminal values.
In the second case, that doesn’t seem to make sense.
In the third case, what we’re discussing is no longer a perfect rational agent.
What exactly do you mean by “perfect rational agent”? Does such a creature exist in reality?