View article

We study here a variant of the alternating-time temporal logic (ATL) where each agent has a given memory. We show that it is an interesting compromise, rather realistic but with a reasonable complexity. In contrast, most models with perfect recall and imperfect information have an undecidable model-checking problem.