Investigating the temporal order of regulatory processes can explain in more detail the mechanisms behind success or lack of success during collaborative learning. The aim of this study is to explore the differences between high- and low-challenge collaborative learning sessions. This is achieved through examining how the three phases of self-regulated learning occur in a collaborative setting and the types of interaction associated with these phases. The participants were teacher training students (N = 44), who worked in groups on a complex task related to didactics of mathematics during 6 weeks. The participants were instructed to use an application that was designed to increase awareness of the cognitive, motivational and emotional challenges the group might face. Based on the application's log files, the sessions were categorized into low- and high-challenge sessions. The video data from each session were coded based on the self-regulation phases and the types of interaction. The frequencies of the phases and the types of interaction were calculated for each session, and process discovery methods were applied using the heuristic miner algorithm. The results show no significant differences between the sessions in the frequency of phases. However, the process models of the two sessions were different: in the high-challenge sessions, the groups switched between the forethought and performance phases more. In conclusion, the regulation phases and types of interaction that contribute to successful collaboration differ in high- and low challenge sessions and support for regulated learning is needed especially at the middle of the learning process.
