In multiagent systems, agents may be led to ask themselves with whom to cooperate, knowing that each of them expresses its own preferences. This problem is studied in hedonic games with solution concepts characterizing the stability of outcomes with respect to the agents' preferences. However, this framework considers a single a priori about agents' common beha-viour. For instance, Nash stability models agents which all want to join the coalitions they prefer without any considerations about the others. Thus, it might also interesting to consider agents which are heterogeneous in their definition of stable solutions. For this purpose, we propose two news hedonic game models. The first one where agents decide the solution concept that they follow, the second one where agents express preferences on the coalitions and the solution concepts. © 2018 Lavoisier.