Retroactive learning

Retroactive learning : reviewing experiences and learning from them when sufficient time (or another resource) becomes available.

Often, it is not possible to learn while an event is occurring because the agent lacks the specific information or resources that it needs to learn. For example, an agent in a real-time environment may not have time to apply an iterative learning algorithm while it is performing a task. However, when time becomes available, the agent can replay the events and learn from them then. Episodic memory allows previous experiences to be relived or rehearsed once the resources are available so it can be reanalyzed with new knowledge or additional experiences.