# Why do we use the Target Network for action evaluation in Double deep Q networks

Is there any specific reason as to why The target Network is used for evaluation and The online network Is used for selection, what would be the difference if both roles were switched, our online network for Evaluation and our target network for selection?