In this study, we explored the extent to which problems and instructional strategies affect social cohesion and interactions for information seeking in physics classrooms. Three sections of a mechanics physics course taught at a Chilean University in Coquimbo were investigated. Each section had a weekly problem-solving session using different sets of well and/or ill-structured problems (i.e., algebra-based and open-ended problems respectively), as well as instructional strategies for guiding the problem-solving sessions. Data was collected on networks of information seeking and perceptions of good physics students, during a problem-solving session. We used social network analysis (SNA) for constructing variables while conducting the study. Results suggest that the teaching and learning strategies to guide problem-solving of well and ill-structured problems yield different levels of social interaction among classmates, and significant levels of activity in seeking out information for learning and problem-solving. While strategies for guiding problem-solving lend to significant differences for network connectivity, well and ill-structured physics problems predict similar levels of social activity.