[madacoo]

It's definitely an interesting question as to whether goal-choice itself can be a rational activity. Of course, it is necessary for a rational agent to begin with axioms and from there make choices. It seems perfectly coherent to me that we could, from those axioms, choose a further set of goals, each of which can be attained following a course of actions, and so on.

And so I would contend that the goal 'cause the cessation of a particular belief' is not axiomatic at all, but an instance of a goal that serves something more fundamentally axiomatic. Like 'make the world more comfortable'. Or 'make the world conform to my expectations'. Or something like that.

From this model we could absolutely talk about choosing goals in a rational manner that aids certain axioms. The question then would be: could the axioms themselves be chosen in a rational manner? Probably not but then it doesn't seem impossible to argue that some sets of axioms are better than others. It would be nice, say, if the axioms guiding the behaviour of a rational agent, didn't contradict each other.

[GoblinSlayer]

Goal choice is adaptation activity. For humans two approaches to adaptation were identified: social and empiric. Monkeys use them too (fun fact: monkeys are more likely to resort to empiric adaptation than humans).