Being There: Why a Robot in the Room Beats the Same Robot on a Screen

Dozens of experiments suggest that physical presence changes how people respond to a social machine, though the effect is more specific, and more interesting, than robots simply being better than screens.

The screen question

Here is an awkward question for anyone selling a social robot. If the clever part is software, the speech, dialogue and personality, why bother with an expensive body at all? A tablet with an animated character can talk, listen and pull faces. It does not need motors, it does not fall over and it costs a fraction of the price. A robot has to justify every gram of hardware.

Human-robot interaction researchers have been testing this for about two decades, typically by comparing the same behaviour delivered in different forms: a robot in the room, the same robot shown live on a screen, and a virtual character on a display. The answers are surprisingly consistent, but they need careful reading.

Early experiments

One influential line of work came from Maja Matarić's group at the University of Southern California. In a 2007 paper, Joshua Wainer, David Feil-Seifer, Dylan Shell and Matarić used a structured Towers of Hanoi task to compare a physically present robot with a remote robot seen through a video link and with a simulated robot on a screen. People rated the physically embodied robot as more appealing and more aware of the world, and as more helpful, watchful and enjoyable than its remote and simulated counterparts. The group's interest was not academic curiosity: they were working on socially assistive robots for people such as children with autism, where engagement is the whole point.

At Yale, Wilma Bainbridge, Justin Hart, Elizabeth Kim and Brian Scassellati compared a humanoid robot that was in the room with the same robot appearing on a live video feed, published in 2011 in the International Journal of Social Robotics. Participants did a book-moving task under the robot's direction. Basic social behaviour, such as greeting the robot and cooperating with simple instructions, looked similar in both conditions. The differences came at the edges. When the robot made an unusual request, people were more likely to go along with it when it was physically present. They also gave the physical robot more personal space, treating it more like a body in their environment. Adding 3D information to the video robot's gestures did not close the gap. Questionnaires matched the behaviour: overall, people had a more positive experience with the robot in the room.

What thirty-odd studies say together

Individual studies are small and their setups vary, so the most useful contribution is Jamy Li's 2015 survey in the International Journal of Human-Computer Studies, which pulled together 33 experimental works comparing copresent robots, telepresent robots shown on screens, and virtual agents.

Li drew a distinction that is easy to miss. A robot can be physically embodied, meaning it has a real body somewhere, without being physically present with you. A live video of a real robot is embodied but not present. A virtual agent is neither. When Li sorted the findings this way, the advantage seemed to come from presence rather than from embodiment as such. Robots in the same room as the user were generally more persuasive and were perceived more positively than robots on screens. But when a real robot was shown on video, people did not respond to it much differently from an animated character. Having a body somewhere else buys you little. Sharing the room is what counts.

That conclusion is both encouraging and humbling for roboticists. It supports the case for physical robots, but it suggests that the value lies less in mechanical sophistication than in occupying the same space as the person, with all the cues of size, distance, direction and potential touch that this brings.

Why presence might matter

Several mechanisms are plausible, though the studies rarely separate them cleanly. A present robot exists in the person's physical space, so its gaze and gestures point at real things. This links to the Mona Lisa problem: on a screen, direction is ambiguous, whereas a body in the room can unmistakably turn towards you or towards an object. Presence also raises stakes. A machine that could bump into you, hand you something or block your path demands a different kind of attention than a picture. And there is novelty, which flatters any physical robot in a short lab study and may wear off over weeks.

There are also good reasons to be cautious. Many studies are brief, use university students and measure attitudes rather than outcomes. Later research has not always found the same advantage, and the size of the effect depends on the task. For information delivery, a screen may serve perfectly well. Where engagement, persuasion or coordination in shared space matters, such as coaching, tutoring, rehabilitation or group conversation, the evidence leans towards physical presence.

The design takeaway

For builders, the lesson is to be clear about what a body is for. If a robot's physical form only hosts a screen with a face, it may capture some of the benefit of presence simply by being there, but it squanders much of what presence enables. The benefits seem to flow from real direction, real proximity and real attention in a shared space. Hybrid designs such as back-projected faces on physical heads, of the kind KTH developed with Furhat, try to get the best of both: the flexibility of animation with the spatial honesty of an object in the room.

The broader point is that social intelligence is not only a matter of software. Some of it lives in where you are standing.

References

1. Li, J. (2015). The benefit of being physically present: A survey of experimental works comparing copresent robots, telepresent robots and virtual agents. International Journal of Human-Computer Studies, 77, 23-37. Link
2. Wainer, J., Feil-Seifer, D. J., Shell, D. A., & Matarić, M. J. (2007). Embodiment and human-robot interaction: A task-based perspective. Proceedings of IEEE RO-MAN 2007, 872-877. Link
3. Bainbridge, W. A., Hart, J. W., Kim, E. S., & Scassellati, B. (2011). The benefits of interactions with physically present robots over video-displayed agents. International Journal of Social Robotics, 3(1), 41-52. Link
4. Al Moubayed, S., Beskow, J., Skantze, G., & Granström, B. (2012). Furhat: A back-projected human-like robot head for multiparty human-machine interaction. In Cognitive Behavioural Systems, Lecture Notes in Computer Science 7403, 114-130. Springer. Link

Feature drafted with AI from the published research listed above, for social-robotics.com.