To simplify our explanation, we only considered only non-associative tasks, in which there is no need to associate different actions with different situations. In a general reinforcement learning task there is more than one situation, and the goal is to learn a policy: a mapping from situations to the actions that are best in those situations.

Associative search task involves both trial-and-error learning in the form of search for the best actions and association of these actions with the situations in which they are best.

They are like the full reinforcement learning problem in that they involve learning a policy, but like our version of the n-armed bandit problem in that each action affect the next situation as well as the reward, then we have the full reinforcement learning problem.