The proliferation of remote laboratory instruction increased substantially during the recent global pandemic. Many physics departments implemented this rapid transition without the previous experience, time, and deliberation to optimize learning experiences for students. The present quasi-experimental, nonequivalent group design study examined social connections and communities of practice in both in-person and remote undergraduate physics laboratories during the Fall 2020 academic semester. A social networking and communities of practice theoretical framework guided the study design and methodology. Study participants (N = 697)included in-person and remote undergraduate students in introductory physics laboratories at a research university in the Northeastern United States. A survey instrument was designed to measure students' perspectives relating to their social connectedness with peers and instructors as well as their physics laboratory self-efficacy. Survey factor analysis identified subdimensions related to student-student social learning perspectives, student-instructor social learning perspectives, and physics laboratory self-efficacy. Analysis of variance indicated remote students experienced weaker levels of engagement with instructors and peers than in-person students, and remote students who connected with one another experienced more social engagement than remote students who did not. Remote students who connected with one another reported having a lower physics laboratory self-efficacy than their in-person counterparts. Isolated remote students did not show a statistical difference in their physics laboratory self-efficacy from their remote-connecting nor in-person counterparts. Correlations between factors were tested, with instructor interactions most closely related to self-efficacy formation. Results suggest that remote laboratories may need formalized mechanisms and incentives to promote social interaction and foster community.