Human Response to Humanoid Robot That Responds to Social Touch

Round 1

Reviewer 1 Report

This is a very interesting study that was inventive and thorough in its experimental design.  I thought the WoZ approach was a sensible choice for this scenario, albeit with reservations (see below). I hope the work is published after revisions, and that future work is possible.  I have some questions regarding both the methods and the results, which both need clearer explanations. 

Methods: 

1.The way the experiment was conducted is not clear.   For example,  I don't understand why the robot responses were random (if that is the case) and how this would help participants make an evaluation of their feelings about the robot's reactions.  Were the participants able to to recall the robot's reactions in every case after the session, when they completed the questionnaire? How did the team match people's responses to whether robot performed positively, negatively or neutrally? Was this done by analysing video in every case (3 sessions with 4 situations each x 40 = 480)?  

2.  We need a better explanation of the "wizard" role, performed by someone watching videos of participants.  How was it possible for that person not to be influenced by what they were viewing?

Presentation:

3. The diagrams provided are very useful and help to provide an overview of the results. I suggest that more visual data would be helpful to explain the text, such as stills from video showing touch happening between robot and human, and a diagram to indicate the patterns of touch used to express different emotions when participants were given freedom to choose. Moreover, a simple flow diagram encapsulating the main activities including how the data was analysed would be helpful.

4. The lengthy descriptions of results under "4. Discussion" are very hard to follow.  It would be better to encapsulate the data in tables or figures and use your text to actually discuss what what was happening.

Results:  

5. Authors note some likely reasons for participants' behaviour in specific cases, such as using methods of touch they had previously been instructed to use, even when given free choice.  Perhaps this implies that the second analysis of touch study (which was theoretically very interesting) deserves more thought in regard to experiment design, since so many factors could have influenced participants' behaviour. E.g., perhaps they believed they had limited options.  Because of this, I would be wary of offering the results of this study as part of the same paper.  Perhaps the question of how humans choose to express emotions to a small robot needs to be a different piece of research, and can be expressed in this paper as ideas for future work.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript shows a series of experiments that assess the subjective feeling of a group of participants that interacts with a humanoid robot by touching it (using affective touch). The robot can have no reaction or different reactions to the affective touch: reactions that are in accordance with the type of touch that was performed on the robot (appropriate reaction) or in disagreement (inappropriate). The reaction was provided by commanding the robot in remote control. Participants tended to have more positive feelings with robots that showed appropriate reactions to the touch of the participants and when the participant was free to interact with the robot to provide different types of feelings. The theme of the manuscript is in line with the thematic of the special issue. The aim of the manuscript is clear and the introduction seems to be well structured by introducing the different types of developed social robots and the performed studies about the importance of affective touch in human-human and human-robot interaction. However, methods, results and discussions sessions are a little confusing and in addition some aspects of the experiment should be better discussed. See the points below.

Points

In Introduction sentence “This phenomenon is called “Midas touch” after the Greek god who turned anything he touched into gold” Midas was not a god but a king of Phrygia in the Greek myth.

It is not clear how the authors selected the affective touch in designed social touch, except for stroking, and also the movement of the robot for appropriate and no appropriate response. Did they follow any previous reference?

The sentence “which gives it an adorable appearance” should be changed in “which makes it judge as adorable”. In addition, the sentence “the participants would not feel uncomfortable stroking its head” sounds like a personal opinion.

In Methods “Place and manner of touch” the authors should better clarify how the data relative to the touch on the robot was recorded. In the analysis part, it seems that they employed video recordings, some data about the material and setup should be added (e.g., camera type, points of placement of camera, how the analysis of the video was performed)

Why did the authors select to employ a non-parametric test to analyze the data? Did the authors used a test to verify the data distribution or did the authors prefer to use such kind of analysis because they are questionnaire data? Please specify in the manuscript

In discussion it seems that the authors split the questions in different categories, a table that summarizes those groups could improve the readability of the manuscript.

In first and second paragraph of “4.1 Results of participant’s responses to questionnaire after each session of designed social touch” session in Discussion, the authors stated that “the scores of the contradictory reactions competed with those of the appropriate reactions for the following four items: “machine-like human-like,” “non-self-aware - self-aware,” “unrefined movement - sophisticated movement,” and “foolish - clever.” Because the robot showed reactive behavior for both appropriate and contradictory reactions, it was considered that there was no difference in the smoothness of movement between the two reactions. … The items that showed no significant difference were “machine-like - human-like,” “non-self-aware - self-aware,” “unrefined movement - sophisticated movement,” and “foolish - clever,” which were items that evaluated the robot’s humanity and intelligence. These results suggest that the participants’ perception of the robot’s animal qualities did not change significantly when the robot showed the appropriate and contradictory reactions. The fact that the robot clearly asserted itself in response to the participant’s touch with a contradictory reaction such as “I wish you would stop” and “I’m scared” may have led to the perception that the robot is a self-aware entity”, providing two different reasons for the same results on the same items, this is confusing. In particular, the first explanation about the smoothness of movement is unclear.

In last paragraph of “4.2 Summary Survey Questionnaire Results” session in Discussion, the authors stated that “if the first session was “appropriate reaction” or “contradictory reaction,” no significant differences were noted among the three sessions”, it seems that a statistical analysis was performed on such kind of data, could the authors explain if such analysis was performed, and which kind of test was employed.

In the sentence “The mean scores for both “emotions were conveyed to NAO” and “NAO understood the emotions” were below the midpoint of the scale (4 points), indicating that the overall perception of the participants regarding the robot was that it did not understand the participants’ emotions. It is possible that the participants expected some kind of response from the robot even when they followed the instructions and performed the intended social touch.”, it is not clear what the authors intend for some kind of response; some examples could help.

In figure 4 the axes and legend labels are too little and difficult to read

The dimension and the appearance of the NAO robot could affect the responses of the evaluated group of participants, this should be discussed in the limitations, if we want to transfer the obtained results to other robot or devices.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thank you for your work addressing the review comments.  The methods and results are now much easier to understand.  I hope you are able to continue with your research.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Review on Applied Sciences

Title: Human Response to Humanoid Robot that Responds to Social Touch

The manuscript shows a series of experiments that assess the subjective feeling of a group of participant that interacts with a humanoid robot by touching it (using affective touch). After the first round of revision, the manuscript resulted to be improved. The authors responded to the main part of concerns of the reviewers. However, some part of the manuscript could be improved:

·       The word “decreasees” at pag. 3 row 100 should be “decreases”

·       The word “frames” could be employed instead of “stills”

·       The type of camera employed to record the participant interaction of the participants with robot should be added (frame rate and camera resolution are important information)

·       For sake of clarity, the session “2.6.2. Survey questionnaire after all sessions” should be “2.6.2. Survey questionnaire after all sessions with interaction with the humanoid robot through designed social touch”, “After the three sessions with NAO” should be “After the three sessions with interaction with NAO through designed social touch”

Author Response

Please see the attachment.

Author Response File: Author Response.pdf