Understanding social interactions among students comprises a rich area of physics education research. Here we focus on the social interactions in introductory physics laboratories (labs). Most existing research in such contexts focuses on within-group social dynamics, however, we argue that interactions between different lab groups are just as valuable, especially in nontraditional (reformed) labs where students have more control over the experimental designs. Using video recordings of lab sessions and social network analysis methods, we examined whether and for how long groups interact with each other, comparing across three offerings of an introductory mechanics course. We observed significant variability in the interactions across lab sessions and no clear pattern in how strongly lab groups were connected in the networks between the different courses. More prolonged intergroup interactions, however, occurred in three of the four reformed lab sections, as compared to a traditional lab section and a reformed lab section that took place in the evening. We also developed a group-level social role classification scheme based on groups' interaction patterns, identifying four roles: noninteractors, information seekers, responders, and mutual interactors. We found that groups in the traditional lab section disproportionately acted as noninteractors; many groups in this lab did not engage in intergroup interactions at all. In contrast, groups in reformed lab sections took on the remaining three roles more frequently. We also found possible relationships between the distributions of the social roles within each type of lab by the groups' gender compositions. All-male groups took on interactive roles more often than all-female and mixed-gender groups in the reformed lab sections, but not in the traditional lab. Results indicate that the amount of time students spend on intergroup interactions may depend on the lab curriculum, the time of day, and student populations.