With the higher education reform putting forward the professionalization of doctoral students, doctoral education has been strongly focused on generic transferable skills to ensure employability. However, doctoral training should not forget core skills of research and especially the ability to formulate research questions, which are the key to original research and difficult to develop at the same time. Learning how to develop a research question is traditionally seen as a one-to-one learning process and an informal daily transmission between a novice and a senior researcher. The objective of this paper is to offer a framework to design doctoral programs aimed at supporting the process of development of research questions for doctoral candidates guided by their supervisors. We base our proposal on two doctoral training programs designed with a pedagogical strategy based on dialogs with peers, whether they be students, supervisors, or trainers from a diversity of scientific backgrounds. The resulting framework combines three learning challenges faced by doctoral students and their supervisors when developing their research question, as well as training objectives corresponding to what they should learn and that are illustrated by the scaffolds we have used in our training programs. Finally, we discuss the conditions and originality of our pedagogical strategy based on the acquisition of argumentation skills, taking both the subjective dimensions of PhD work and the added value of interactions with a diversity and heterogeneity of peers into account.