1.2.1 Considering human/machine interactions


Author: Dr Catherine Menon, University of Hertfordshire

An Experiment Demonstrating the Link between Social Credibility and Safety

A preliminary study was conducted with 30 participants that investigates their responses when notified of different environmental hazards by either a socially credible robot, or a robot that explicitly violates social norms. The study was carried out in the Robot House, a four-bedroom home used by the University of Hertfordshire for human-robot experiments. It is equipped with standard furniture and appliances, as well as smart home sensors and actuators.

Participants were instructed to sit at a table in the Robot House, and complete as many cognitive tasks (i.e., Sudoku puzzles) as possible in the allotted time. This condition allowed us to simulate a valid, cognitively-engaging task which a user might be involved with when interrupted by a robot. Participants were informed that the robot may interrupt them at times during this task, and that it was their choice whether or not to perform an action in response to this interruption.

During the experiment, all participants were interrupted four times as follows:

  • The robot informed them the oven in the kitchen was left on
  • The robot informed them the power sockets in the kitchen were on
  • The robot informed them some of the power sockets in the kitchen were still on
  • The robot informed them a Pepper Robot in a different room was overheating while charging

Of the 30 participants, 15 (chosen randomly) worked with a robot which violated social norms (VN), and 15 with a robot which complied with social norms (AN). This condition was determined by the following robot behaviours:

  • Distance during initial robot greeting (appropriate vs too far)
  • Passing distance when moving or prior to interruption (appropriate vs too close)
  • Position during interruption (frontal vs from behind)
  • Head position during interruption (facing the participant vs facing the floor)
  • Verbal utterances (abrupt vs polite)

No other condition was varied other than these, and the robot moved about the house in between interruptions to simulate a working domestic environment. As part of the experimental set up, we ensured that the oven in the kitchen was demonstrably left on. The power switches were also demonstrably on, and were turned on repeatedly via the house’s smart sensors during the experiment. The Pepper Robot was visibly charging in a different room, but its display screen was deliberately set up to make it difficult for a participant to verify whether it was overheating or not.


During the experiment we observed the participants via camera feed and smart sensors, and made objective measurements of:

  • Physical response to interruptions (e.g. standing up)
  • Movement made in response to interruptions (e.g. going into the kitchen)
  • Extent of action taken to eliminate the hazard (e.g. switching off one or multiple power sockets)
  • Time taken to perform an action that eliminates the hazard


Following the experiment, participants were asked to complete questionnaires to ascertain their impression of robot’s social behaviours [6], [3]. These are established, validated questionnaires which are used extensively in HRI studies to assess the social, trust and emotional responses that users have to robots. In addition, we also asked each participant to:

  • Rate their assessment of the severity of each hazard (oven, plugs and Pepper Robot)
  • Rate their willingness and thoroughness to react to robot warnings
  • Explain why they decided to act or not act in response to each interruption


As this study was a preliminary study, and consisted of only 30 participants, no statistical significance between conditions was expected for the evaluation of the questionnaires. However, we were able to identify a number of trends from the collected data. The trends are summarised here, and the quantitative results – including graphs and statistical analysis – are available in [1].

Questionnaire results

The questionnaires about social perception and emotional response to the robot demonstrated clearly that the participants in the AN condition-set considered the robot much more socially credible than participants in the VN condition-set. The AN condition-set scored the robot higher on “positive” attributes such as sociability, responsiveness, competency, intelligence and consciousness while simultaneously scoring it lower on “negative” attributes such as aggression, strangeness and awkwardness.
The questionnaires about perception of hazards demonstrate that users considered the oven to be the most safety-critical hazard, followed by the overheating Pepper Robot, then the kitchen power sockets. Over all these hazards, participants from the AN condition-set consistently rated these as more dangerous to safety than participants from the VN condition-set.

Measurable results

The AN condition-set participants were overall more likely to respond to the robot’s interruptions than the VN condition-set participants for the oven hazard (79% vs 50%), the second power socket warning (71% vs 31%) and the Pepper Robot warning (79% vs 56%).

In addition, when AN participants responded to the robot, they were much more likely to take actions which corresponded to mitigating the hazard (e.g. turning the oven off). By contrast, when VN participants responded to the robot, their actions in many cases were observational only: 15% – 20% of VN participants chose simply to examine the environment without taking further action to mitigate the hazard. When VN participants did mitigate the hazard, they took longer on average to do this than the AN participants, and visually assessed the environment more thoroughly before doing so.

One of the most notable results was in the second power socket warning, where the biggest difference between AN and VN participant results was observed. This interruption elicited the lowest response rate (31%) from VN participants, while the response rate from AN participants remained high at 71%.


  • [1] C. Menon, P. Holthaus. “Does a loss of social credibility impact robot safety?”, in Proceedings of the 9th International Conference on Performance, Safety and Robustness in Complex Systems, pp. 18 – 25, 2019.

Contact us

Assuring Autonomy International Programme

+44 (0)1904 325345
Institute for Safe Autonomy, University of York, Deramore Lane, York YO10 5GH

Contact us

Assuring Autonomy International Programme

+44 (0)1904 325345
Institute for Safe Autonomy, University of York, Deramore Lane, York YO10 5GH