A video-conferencing system is described that provides physical cues regarding remote participants. Each remote participant is physically represented at a video conference by a robotic unit that includes a monitor, camera, microphone and speaker. In this way, a physical presence of the remote participant is conveyed at the conference, so that other participants are more likely to involve the remote participant. Moreover, the remote participant has access to a gesture determination system, which inputs gesture information about the remote participant that expresses a state of mind of the participant. Such gesture information may include leaning forward to show interest, or leaning back to show disinterest. The gesture information is transmitted to the robotic unit, which is actuated so as to reflect the gesture information, and thereby express the state of mind of the remote participant in a physical, intuitive way.