In face-to-face interactions, people rely on facial expressions, body posture, and further verbal and non-verbal cues to infer others’ internal states. While collaborative virtual reality (VR) enables rich shared experiences, it offers fewer expressive cues, making perceiving mental states like attention, intention, or cognitive load (CL) more difficult. Rather than replicating real-world cues, we investigated how augmenting collaborative VR with real-time visualizations of users’ CL impacts social presence. We first conducted semi-structured interviews (N = 6) and an online survey (N = 54) to elicit user needs and design preferences for CL visualizations. These insights informed the development of a collaborative VR application evaluated in a lab experiment (N = 30). Participants engaged in verbal explanation tasks of varying difficulty while their CL was displayed using different visualizations (pie charts, bar charts, brain diagrams), including a no-visualization control condition. Social presence was measured using the Networked Minds Social Presence Inventory and augmented via post-experiment interviews. Our results show that in difficult tasks, pie chart visualizations helped sustain co-presence, while all diagrams supported affective understanding. These findings demonstrate the potential of intuitive CL visualizations to foster social connectedness, helping inform future designs of collaborative VR.