Gaze-Based Interaction Intention Recognition in Virtual Reality
Round 1
Reviewer 1 Report
See attached file.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
This is a nice piece of work on exploring hand-eye coordination-related features to improve interaction intention recognition in a virtual reality environment.
Author Response
Thanks for taking the time to review our article. We appreciate your recognition of our work.
Reviewer 3 Report
This paper provides information about virtual reality visualizations and eye-tracking technology in the context of predicting users’ interactions intention. Presented results give insight into the future of predictive interfaces.
In terms of related work, I would suggest citing “Zagata K., Gulij J., Halik Ł., Medyńska-Gulij B., 2021, Mini-Map for Gamers Who Walk and Teleport in a Virtual Stronghold, ISPRS Int. J. Geo-Inf. 2021, 10, 96, DOI: 10.3390/ijgi10020096.”. This research, I believe is related to your studies (in terms of possible applications), especially that mentioned paper refers to different techniques of location change (including teleportation and picking items in VR) within the game scenario including eye movement technology.
You state that your study participants had normal or corrected-to-normal vision. Does it mean they wore glasses? Please add more detail on this.
Your study refers to hand-eye coordination. However, you have only one feature (the Hit Point of the Controller Ray, called ‘distance’ later on) related to controller movement, and the rest are derived from eye movement. Therefore, I am not fully convinced that your study refers to hand-eye interaction. Even your results strongly suggest that eye movement features play a crucial role in prediction. In my opinion, it would be more appropriate to track the users’ controller pad all the time. I think that it would be even a better feature for prediction.
In table 4 what are C, F, and S?
You have slightly discussed limitations. In my opinion, the main limitation is that you did not explicitly define why you choose these types of interactions. Right now, the potential reader could have the impression that this is art for art’s sake. Could you detail examples of the usage of this type of interaction such as teleportation or picking items from a distance? But please, do not say that “it would have an application in VR” because it's too general. Please show how your interaction types may be applicable in real scenarios.
Another limitation is what you have described as “not entirely naturalistic”. It would be necessary, in further studies, to test your classification performance in a more complicated situation. For example including different tasks such as identifying objects, walking, moving items, etc. It would be interesting for potential readers to present in more detail the possibilities of the classification predictors. And the question arises: how many different tasks could be correctly recognized using only these types of features?
Also, the environmental aspect does not been discussed. Teleportation could be predicted when there is a free space to move in. This interaction could be highly precise by a learning model without a user’s controller. Picking things require items.
One more thing to add. Implementation of an intuitive interface based on prediction requires not only specifying the correct classification of the data for proper task. It is also about the user's decision to perform the interaction. Participants often have multiple objects to interact with, but they may choose only one of them or neither. Speaking of implementation, future studies would need to incorporate what features of participants' behavior are crucial for the decision to enter the interaction with the specific item. Please discuss the possibilities of classifying the decision process between the same task.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The paper notably improved after its revision. All of my comments were properly addressed, and all of my doubts were resolved. Well done.
