the dynamics of emotions in online interaction - open...

© 2016 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution

License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author

and source are credited

Review History

RSOS-150214.R0 (Original submission)

Review form: Reviewer 1 (Jeff Hancock) Is the manuscript scientifically sound in its present form? No Are the interpretations and conclusions justified by the results? No Is the language acceptable? Yes Is it clear how to access all supporting data? I could not find any indication of how the authors planned to share their data. Do you have any ethical concerns with this paper? No Have you any concerns about statistical analyses in this paper?

The dynamics of emotions in online interaction

David Garcia, Arvid Kappas, Dennis Küster and Frank Schweitzer

Article citation details

R. Soc. open sci. 3: 160059.

http://dx.doi.org/10.1098/rsos.160059

Review timeline

Original submission: 19 May 2015

1st revised submission: 27 January 2016

2nd revised submission: 12 May 2016

3rd revised submission: 11 July 2016

Final acceptance: 12 July 2016

Note: Reports are unedited and appear as submitted

by the referee. The review history appears in

chronological order.

on August 17, 2018http://rsos.royalsocietypublishing.org/Downloaded from

http://rsos.royalsocietypublishing.org/

2

Yes Recommendation? Major revision is needed (please make suggestions in comments) Comments to the Author(s) The paper “The Dynamics of Emotions in Online Interactions” presents three studies that examine emotional dynamics after reading or producing content in an online discussion forum. The studies examine changes in valence and arousal from self-report measures, finding evidence for, among other things, 1) a relaxation towards baseline 2) changes in valence that are dependent on the content of the discussion, and 3) when producing posts for an online discussion arousal decreases as part of a regulation mechanism. I was excited to read this paper as I think the topic is important and the approach the authors were taking was interesting. There is indeed much to like about the paper. It takes on important questions at the micro level of emotion rather than more global questions, which is more typical. In particular, the dynamic temporal nature of these analyses is valuable. The statistical analyses were sophisticated and the figures were clear. Unfortunately, I have two concerns that undermine the contribution of the paper, one a methodological issue and the other a framing issue. The most important concern is that all of the data in the study, across the 3 experiments, are all based on a single-item measure: a Likert-scale item from 1-7 that the authors transform in to valence and arousal values. First, the transformation is not made clear, but more importantly, most (perhaps all) theories of emotion posit two distinct dimensions of emotions, as does the current paper. For example, Russell’s single-item measure of emotion (1989) involved a two dimensional grid that could capture both dimensions. More generally, single-item measures tend to be viewed skeptically given their poor psychometric qualities. Given the very sophisticated statistics used here, I am concerned that the quality of the underlying data is not sufficient to be reliable for these analyses. For me to be able to recommend this paper, I would need to be convinced that this kind of single-item measure for a multi-dimensional construct is sufficiently reliable to support these analyses, either from citations in the literature or gathering new data with more appropriate measures. Otherwise, its really not possible to make reasonable conclusions from the analyses.The same holds for the likelihood to participate - while this is likely a single dimension, its hard to tell since there is only one item, the general psychometric weakness of a single-item measure applies. The second issue is the framing of the study as modeling the dynamics of emotions in online interactions. While this is descriptive of the stimuli, the current model and empirical test make no indication of how this is different from any other stimulus. For instance, is there something special about reading some content in an online discussion forum, than say, a newspaper, or a diary, or a novel? What is it that is unique to the online context that suggests that this model and these data are specific to the dynamics of online interaction? This criticism is obvious for studies 1 and 2 given that there is no interaction either, but for study 3 I think the authors could make a good case. The participants did in fact interact with the context and provided their own messages. To me, this interactivity is an important feature of the online environment, though not the only one. The paper fails to indicate why this is about online interaction and not a more general model (which might be fine to articulate). If these issues can be resolved, I would be happy to review a revision and provide more detailed, lower-level comments.



3

Review form: Reviewer 2 (Kathy Robinson) Is the manuscript scientifically sound in its present form? No Are the interpretations and conclusions justified by the results? No Is the language acceptable? Yes Is it clear how to access all supporting data? Information about the threads used for Studies 1 & 2 would be helpful. What topics did they cover? How many posts in each thread? Can you provide some detail about the functionality of the threading technology (for example, can participants in the forum use selective quoting) Please provide the questions given to research participants and provide more detail about how the participant responses were transformed into valence and arousal levels. Also, please provide a tabulated summary of participants' subjective assessments. Why did Study 2 involve a subset of threads? how was the subset chosen? Study 3: Could you provide some detail about how subjective assessment was translated into a form that could be used by the agent. Do you have any ethical concerns with this paper? No Have you any concerns about statistical analyses in this paper? I do not feel qualified to assess the statistics Recommendation? Major revision is needed (please make suggestions in comments) Comments to the Author(s) Introduction I read the Introduction more than once yet remained unclear about the research framework on which this article is based; the reader is not led from the general to the specific. You refer to studies of collective phenomena yet do not situate you research within a specific theory i.e. identification, suggestibility, the uncanny, patterning of affective practice, are some that have been proposed by other researchers ( you do mention emotional contagion, briefly). Yet at l51 p1 you imply that an agent is capable of 'unifying different models of social interaction'. Identification during online social interaction online has been extensively researched and is likely to influence the dialogue on which your study is based ( inferred from thread message example supplied in S1). Similarly Wetherell's proposal in Affect and Emotion to study the patterning of affect has synergy with your intention to study the dynamics of online interaction. Am I correct in assuming that you are referring to excitability as a general emotion. Otherwise I cannot understand how forum participants can share a collective emotion given the topic covered in the example message. Emotional agents model I note that in the Discussion you describe the Cybermotions framework as based on assumptions. This needs to be made clear from the outset also, what is the rationale behind these assumptions? Some current work has incorporated a dominance dimension alongside valence and amplitude so you need to explain in more detail why you have not taken individual differences in emotional resilience into account.



4

Results Figure 3 would be more convincing if you had used an an average of all participants Discussion By using an agent and an experimental paradigm you claim that most of the ethical concerns related to studying online social interaction are circumvented. However, what you now have is a first person social interaction for Studies 1 & 2 and is a limitation that should be mentioned. Also, could you provide some detail about the ethical guidelines you used when sourcing the threads and also for ensuring the wellbeing of your participants. Finally, although you sourced the conversation from an online discussion you do not make any reference to what is unique about this particular medium for of interpersonal communication - there is an extensive research literature in related disciplines for example, Communication, Distance Education

Decision letter (RSOS-150214)

20th August 2015 Dear Dr Garcia: Manuscript ID RSOS-150214 entitled "The Dynamics of Emotions in Online Interaction" which you submitted to Royal Society Open Science, has been reviewed. The comments from reviewers are included at the bottom of this letter. In view of the criticisms of the reviewers, the manuscript has been rejected in its current form. However, a new manuscript may be submitted which takes into consideration these comments. Please note that resubmitting your manuscript does not guarantee eventual acceptance, and that your resubmission will be subject to peer review before a decision is made. You will be unable to make your revisions on the originally submitted version of your manuscript. Instead, revise your manuscript and upload the files via your author centre. Once you have revised your manuscript, go to https://mc.manuscriptcentral.com/rsos and login to your Author Center. Click on "Manuscripts with Decisions," and then click on "Create a Resubmission" located next to the manuscript number. Then, follow the steps for resubmitting your manuscript. Your resubmitted manuscript should be submitted by 17-Feb-2016. If you are unable to submit by this date please contact the Editorial Office. We look forward to receiving your resubmission. Sincerely, Emilie Aime Senior Publishing Editor, Royal Society Open Science [email protected] Associate Editor Comments to Author (Dr Jonathan Roiser): Associate Editor: 1 Comments to the Author: Both reviewers have major concerns that need to be addressed.



5

Reviewers' Comments to Author: Reviewer: 1 Comments to the Author(s) The paper “The Dynamics of Emotions in Online Interactions” presents three studies that examine emotional dynamics after reading or producing content in an online discussion forum. The studies examine changes in valence and arousal from self-report measures, finding evidence for, among other things, 1) a relaxation towards baseline 2) changes in valence that are dependent on the content of the discussion, and 3) when producing posts for an online discussion arousal decreases as part of a regulation mechanism. I was excited to read this paper as I think the topic is important and the approach the authors were taking was interesting. There is indeed much to like about the paper. It takes on important questions at the micro level of emotion rather than more global questions, which is more typical. In particular, the dynamic temporal nature of these analyses is valuable. The statistical analyses were sophisticated and the figures were clear. Unfortunately, I have two concerns that undermine the contribution of the paper, one a methodological issue and the other a framing issue. The most important concern is that all of the data in the study, across the 3 experiments, are all based on a single-item measure: a Likert-scale item from 1-7 that the authors transform in to valence and arousal values. First, the transformation is not made clear, but more importantly, most (perhaps all) theories of emotion posit two distinct dimensions of emotions, as does the current paper. For example, Russell’s single-item measure of emotion (1989) involved a two dimensional grid that could capture both dimensions. More generally, single-item measures tend to be viewed skeptically given their poor psychometric qualities. Given the very sophisticated statistics used here, I am concerned that the quality of the underlying data is not sufficient to be reliable for these analyses. For me to be able to recommend this paper, I would need to be convinced that this kind of single-item measure for a multi-dimensional construct is sufficiently reliable to support these analyses, either from citations in the literature or gathering new data with more appropriate measures. Otherwise, its really not possible to make reasonable conclusions from the analyses.The same holds for the likelihood to participate - while this is likely a single dimension, its hard to tell since there is only one item, the general psychometric weakness of a single-item measure applies. The second issue is the framing of the study as modeling the dynamics of emotions in online interactions. While this is descriptive of the stimuli, the current model and empirical test make no indication of how this is different from any other stimulus. For instance, is there something special about reading some content in an online discussion forum, than say, a newspaper, or a diary, or a novel? What is it that is unique to the online context that suggests that this model and these data are specific to the dynamics of online interaction? This criticism is obvious for studies 1 and 2 given that there is no interaction either, but for study 3 I think the authors could make a good case. The participants did in fact interact with the context and provided their own messages. To me, this interactivity is an important feature of the online environment, though not the only one. The paper fails to indicate why this is about online interaction and not a more general model (which might be fine to articulate). If these issues can be resolved, I would be happy to review a revision and provide more detailed, lower-level comments. Reviewer: 2 Comments to the Author(s)



6

Introduction I read the Introduction more than once yet remained unclear about the research framework on which this article is based; the reader is not led from the general to the specific. You refer to studies of collective phenomena yet do not situate you research within a specific theory i.e. identification, suggestibility, the uncanny, patterning of affective practice, are some that have been proposed by other researchers ( you do mention emotional contagion, briefly). Yet at l51 p1 you imply that an agent is capable of 'unifying different models of social interaction'. Identification during online social interaction online has been extensively researched and is likely to influence the dialogue on which your study is based ( inferred from thread message example supplied in S1). Similarly Wetherell's proposal in Affect and Emotion to study the patterning of affect has synergy with your intention to study the dynamics of online interaction. Am I correct in assuming that you are referring to excitability as a general emotion. Otherwise I cannot understand how forum participants can share a collective emotion given the topic covered in the example message. Emotional agents model I note that in the Discussion you describe the Cybermotions framework as based on assumptions. This needs to be made clear from the outset also, what is the rationale behind these assumptions? Some current work has incorporated a dominance dimension alongside valence and amplitude so you need to explain in more detail why you have not taken individual differences in emotional resilience into account. Results Figure 3 would be more convincing if you had used an an average of all participants Discussion By using an agent and an experimental paradigm you claim that most of the ethical concerns related to studying online social interaction are circumvented. However, what you now have is a first person social interaction for Studies 1 & 2 and is a limitation that should be mentioned. Also, could you provide some detail about the ethical guidelines you used when sourcing the threads and also for ensuring the wellbeing of your participants. Finally, although you sourced the conversation from an online discussion you do not make any reference to what is unique about this particular medium for of interpersonal communication - there is an extensive research literature in related disciplines for example, Communication, Distance Education

Author's Response to Decision Letter for (RSOS-150214)

See Appendix A.

RSOS-160059.R0 (Revision)

Review form: Reviewer 2 (Kathy Robinson) Is the manuscript scientifically sound in its present form? No Are the interpretations and conclusions justified by the results? Yes



7

Is the language acceptable? Yes Is it clear how to access all supporting data? Yes Do you have any ethical concerns with this paper? No Have you any concerns about statistical analyses in this paper? I do not feel qualified to assess the statistics Recommendation? Major revision is needed (please make suggestions in comments) Comments to the Author(s) The authors have improved this submission significantly. However, a major problem remains; the assumption that the processes that result in shared emotions online are synonymous with those experienced face-to-face (as expressed at line 14 of the Introduction). There has been considerable debate and associated research on the relational implications of mediated communication and the findings that have relevance for the design of the Cyberemotions modelling framework should be addressed. There is a related issue. The date of publication for reference 33, included as an example of online identification, is 1978 which predates the opportunity to communicate with others through forums and other social media by several years. The report is focused on collective emotions. The new material in the Introduction focuses on group emotions. The psychology of group interpersonal relations does not necessarily translate to collective emotions. The discussion topic as illustrated by one message (S1) appears to be emotive and is why the ethical criteria were requested. The report claims to be following the emotional valence and arousal of the interaction as it evolves. Indeed it is a significant advantage of the approach. However, the account of the ethics in Section 5a implies that the discussion thread was manipulated in some way – see lines 15,16 - that in fact participants did not interact with the discussion as it evolved.

Review form: Reviewer 3 Is the manuscript scientifically sound in its present form? No Are the interpretations and conclusions justified by the results? No Is the language acceptable? Yes Is it clear how to access all supporting data? Yes Do you have any ethical concerns with this paper?



8

No Have you any concerns about statistical analyses in this paper? I do not feel qualified to assess the statistics Recommendation? Accept with minor revision (please list in comments) Comments to the Author(s) This article reports a very interesting project. It aims at reducing the emotional aspects of on-line interactions to basics so as to efficiently test assumptions about the dynamics of emotions in social interactions on the Internet networks. The project raises a major concern of current research on emotion and communication across social networks. It offers a convincing demonstration of the usefulness of the proposed method and concepts. However, this innovative project brings to the floor a number of concepts that readers who study emotions are unfamiliar with. In many places, the text of the article would deserve to be clearer or more explicit in order to avoid losing precisely the readership that might benefit from it. - the notion of "models of individual emotion dynamics" is introduced with a list of examples (p. 3, "cellular automata", "bifurcations in dynamical systems", "Brownian motion"). The reader might appreciate being explained the concept in some details. - A bit further in the text comes the "Cyberemotions modeling framework". In the sentence that follows, it is said that this framework "allows the creation of models of user emotions etc." But besides this definition by the use, nothing is said of what "Cyerberemotions framework" means and represents, where it comes from and so forth The model that is at the core of the paper is described under the title 2 "Emotional agents model". I had a hard time to go through it and to figure out exactly what the variables were. As far as possible, I would prefer reading first a detailed and clarified exposure of the variables in the model and their interrelations, followed in a later step by the set of equation formalizing them. As an illustration of the difficulties I met, "expression s" is first mentioned in the legend of Figure 1. It is only later on p. 4 that the notion of "expression" shows up in the text supposed to describe (articulate!) the model. And it is lost in a sentence stating that "the dynamics of the perceived emotional content (…) are influenced by the emotional content of the expression of the agents, si(t)". The notion of "expression rules" emerges a number of times in the article but as far as I have seen, it is never introduced or defined. As regards the Methods, I regretted that the methods section took place only at the end of the paper. But this might be a specific requirement of the journal. In any case, it is impossible to go through the Results and Discussion sections without a minimal knowledge of what was done in study 1, 2 and 3. So, this minimal information could be sketched before the beginning of Section 3. The reader might want to know the context of the studies (what was said to participants?). Information available about studies 1 and 2 is really minimal, and virtually absent for study 3—which should certainly be corrected. In conclusion, I am clearly in favor of the publication of this innovative and promising work. But I believe that the text of the paper needs to be clarified and complemented in a good deal of respect.



9

Decision letter (RSOS-160059)

22nd April 2016 Dear Dr Garcia, The Subject Editor assigned to your paper ("The Dynamics of Emotions in Online Interaction") has now received comments from reviewers. We would like you to revise your paper in accordance with the referee and Subject Editor suggestions which can be found below (not including confidential reports to the Editor). Please note this decision does not guarantee eventual acceptance. Please submit a copy of your revised paper within three weeks (i.e. by the 14th May 2016). If we do not hear from you within this time then it will be assumed that the paper has been withdrawn. In exceptional circumstances, extensions may be possible if agreed with the Editorial Office in advance. We do not allow multiple rounds of revision so we urge you to make every effort to fully address all of the comments at this stage. If deemed necessary by the Editors, your manuscript will be sent back to one or more of the original reviewers for assessment. If the original reviewers are not available we may invite new reviewers. To revise your manuscript, log into http://mc.manuscriptcentral.com/rsos and enter your Author Centre, where you will find your manuscript title listed under "Manuscripts with Decisions." Under "Actions," click on "Create a Revision." Your manuscript number has been appended to denote a revision. Revise your manuscript and upload a new version through your Author Centre. When submitting your revised manuscript, you must respond to the comments made by the referees and upload a file "Response to Referees" in "Section 6 - File Upload". Please use this to document how you have responded to each of the comments, and the adjustments you have made. In order to expedite the processing of the revised manuscript, please be as specific as possible in your response. In addition to addressing all of the reviewers' and editor's comments please also ensure that your revised manuscript contains the following sections before the reference list: • Ethics statement If your study uses humans or animals please include details of the ethical approval received, including the name of the committee that granted approval. For human studies please also detail whether informed consent was obtained. For field studies on animals please include details of all permissions, licences and/or approvals granted to carry out the fieldwork. • Data accessibility It is a condition of publication that all supporting data are made available either as supplementary information or preferably in a suitable permanent repository. The data accessibility section should state where the article's supporting data can be accessed. This section should also include details, where possible of where to access other relevant research materials such as statistical tools, protocols, software etc can be accessed. If the data has been deposited in an external repository this section should list the database, accession number and link to the DOI for all data from the article that has been made publicly available. Data sets that have been deposited in an external repository and have a DOI should also be appropriately cited in the manuscript and included in the reference list. If you wish to submit your supporting data or code to Dryad (http://datadryad.org/), or modify your current submission to dryad, please use the following link:



10

http://datadryad.org/submit?journalID=RSOS&manu=RSOS-160059 • Competing interests Please declare any financial or non-financial competing interests, or state that you have no competing interests. • Authors’ contributions All submissions, other than those with a single author, must include an Authors’ Contributions section which individually lists the specific contribution of each author. The list of Authors should meet all of the following criteria; 1) substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; 2) drafting the article or revising it critically for important intellectual content; and 3) final approval of the version to be published. All contributors who do not meet all of these criteria should be included in the acknowledgements. We suggest the following format: AB carried out the molecular lab work, participated in data analysis, carried out sequence alignments, participated in the design of the study and drafted the manuscript; CD carried out the statistical analyses; EF collected field data; GH conceived of the study, designed the study, coordinated the study and helped draft the manuscript. All authors gave final approval for publication. • Acknowledgements Please acknowledge anyone who contributed to the study but did not meet the authorship criteria. • Funding statement Please list the source of funding for each author. Once again, thank you for submitting your manuscript to Royal Society Open Science and I look forward to receiving your revision. If you have any questions at all, please do not hesitate to get in touch. Yours sincerely, Andrew Dunn Senior Publishing Editor, Royal Society Open Science [email protected] Comments to Author: Reviewers' Comments to Author: Reviewer: 2 Comments to the Author(s) The authors have improved this submission significantly. However, a major problem remains; the assumption that the processes that result in shared emotions online are synonymous with those experienced face-to-face (as expressed at line 14 of the Introduction). There has been considerable debate and associated research on the relational implications of mediated communication and the findings that have relevance for the design of the Cyberemotions modelling framework should be addressed. There is a related issue. The date of publication for



11

reference 33, included as an example of online identification, is 1978 which predates the opportunity to communicate with others through forums and other social media by several years. The report is focused on collective emotions. The new material in the Introduction focuses on group emotions. The psychology of group interpersonal relations does not necessarily translate to collective emotions. The discussion topic as illustrated by one message (S1) appears to be emotive and is why the ethical criteria were requested. The report claims to be following the emotional valence and arousal of the interaction as it evolves. Indeed it is a significant advantage of the approach. However, the account of the ethics in Section 5a implies that the discussion thread was manipulated in some way – see lines 15,16 - that in fact participants did not interact with the discussion as it evolved. Reviewer: 3 Comments to the Author(s) This article reports a very interesting project. It aims at reducing the emotional aspects of on-line interactions to basics so as to efficiently test assumptions about the dynamics of emotions in social interactions on the Internet networks. The project raises a major concern of current research on emotion and communication across social networks. It offers a convincing demonstration of the usefulness of the proposed method and concepts. However, this innovative project brings to the floor a number of concepts that readers who study emotions are unfamiliar with. In many places, the text of the article would deserve to be clearer or more explicit in order to avoid losing precisely the readership that might benefit from it. - the notion of "models of individual emotion dynamics" is introduced with a list of examples (p. 3, "cellular automata", "bifurcations in dynamical systems", "Brownian motion"). The reader might appreciate being explained the concept in some details. - A bit further in the text comes the "Cyberemotions modeling framework". In the sentence that follows, it is said that this framework "allows the creation of models of user emotions etc." But besides this definition by the use, nothing is said of what "Cyerberemotions framework" means and represents, where it comes from and so forth The model that is at the core of the paper is described under the title 2 "Emotional agents model". I had a hard time to go through it and to figure out exactly what the variables were. As far as possible, I would prefer reading first a detailed and clarified exposure of the variables in the model and their interrelations, followed in a later step by the set of equation formalizing them. As an illustration of the difficulties I met, "expression s" is first mentioned in the legend of Figure 1. It is only later on p. 4 that the notion of "expression" shows up in the text supposed to describe (articulate!) the model. And it is lost in a sentence stating that "the dynamics of the perceived emotional content (…) are influenced by the emotional content of the expression of the agents, si(t)". The notion of "expression rules" emerges a number of times in the article but as far as I have seen, it is never introduced or defined. As regards the Methods, I regretted that the methods section took place only at the end of the paper. But this might be a specific requirement of the journal. In any case, it is impossible to go through the Results and Discussion sections without a minimal knowledge of what was done in study 1, 2 and 3. So, this minimal information could be sketched before the beginning of Section 3. The reader might want to know the context of the studies (what was said to participants?). Information available about studies 1 and 2 is really minimal, and virtually absent for study 3—which should certainly be corrected.



12

In conclusion, I am clearly in favor of the publication of this innovative and promising work. But I believe that the text of the paper needs to be clarified and complemented in a good deal of respect. Associate Editor's comments (Dr Jonathan Roiser): Associate Editor Comments to the Author: Editor's note: Both reviewers agree that the manscript has improved following revision, but still have concerns that need to be addressed.

Author's Response to Decision Letter for (RSOS-160059)

See Appendix B.

RSOS-160059.R1 (Revision)

Review form: Reviewer 2 (Kathy Robinson) Is the manuscript scientifically sound in its present form? Yes Are the interpretations and conclusions justified by the results? Yes Is the language acceptable? Yes Is it clear how to access all supporting data? Yes Do you have any ethical concerns with this paper? No Have you any concerns about statistical analyses in this paper? I do not feel qualified to assess the statistics Recommendation? Accept with minor revision (please list in comments) Comments to the Author(s) The extended experimental design section has clarified most of the issues previously raised. However, the references to group identification remain a concern specifically, that the identification differences between face-to-face and mediated communication in the context of ‘the group’ have not yet been addressed. Also, I note the frequent use of the qualifier ‘likely’ when describing group processes and that there is no discussion of the implication of the findings from the research for group processes. The context described for the submission is an open discussion forum involving large numbers of people and the collective emotions that ensue. Why is



13

reference to ‘the group’ relevant? My preference would be for all reference to group identification to be omitted from the Introduction.

Review form: Reviewer 3 Is the manuscript scientifically sound in its present form? Yes Are the interpretations and conclusions justified by the results? Yes Is the language acceptable? Yes Is it clear how to access all supporting data? Yes Do you have any ethical concerns with this paper? No Have you any concerns about statistical analyses in this paper? No Recommendation? Accept as is Comments to the Author(s) The authors should be congratulated for the efforts they invested in meeting the reviewers' remarks and suggestions and in improving considerably the readability of their manuscript.

Decision letter (RSOS-160059.R1)

2nd July 2016 Dear Dr Garcia: On behalf of the Editors, I am pleased to inform you that your Manuscript RSOS-160059.R1 entitled "The Dynamics of Emotions in Online Interaction" has been accepted for publication in Royal Society Open Science subject to minor revision in accordance with the referee suggestions (please address reviewer 1 concerns). Please find the referees' comments at the end of this email. The reviewers and Subject Editor have recommended publication, but also suggest some minor revisions to your manuscript. Therefore, I invite you to respond to the comments and revise your manuscript. • Ethics statement



14

If your study uses humans or animals please include details of the ethical approval received, including the name of the committee that granted approval. For human studies please also detail whether informed consent was obtained. For field studies on animals please include details of all permissions, licences and/or approvals granted to carry out the fieldwork. • Data accessibility It is a condition of publication that all supporting data are made available either as supplementary information or preferably in a suitable permanent repository. The data accessibility section should state where the article's supporting data can be accessed. This section should also include details, where possible of where to access other relevant research materials such as statistical tools, protocols, software etc can be accessed. If the data has been deposited in an external repository this section should list the database, accession number and link to the DOI for all data from the article that has been made publicly available. Data sets that have been deposited in an external repository and have a DOI should also be appropriately cited in the manuscript and included in the reference list. If you wish to submit your supporting data or code to Dryad (http://datadryad.org/), or modify your current submission to dryad, please use the following link: http://datadryad.org/submit?journalID=RSOS&manu=RSOS-160059.R1 • Competing interests Please declare any financial or non-financial competing interests, or state that you have no competing interests. • Authors’ contributions All submissions, other than those with a single author, must include an Authors’ Contributions section which individually lists the specific contribution of each author. The list of Authors should meet all of the following criteria; 1) substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; 2) drafting the article or revising it critically for important intellectual content; and 3) final approval of the version to be published. All contributors who do not meet all of these criteria should be included in the acknowledgements. We suggest the following format: AB carried out the molecular lab work, participated in data analysis, carried out sequence alignments, participated in the design of the study and drafted the manuscript; CD carried out the statistical analyses; EF collected field data; GH conceived of the study, designed the study, coordinated the study and helped draft the manuscript. All authors gave final approval for publication. • Acknowledgements Please acknowledge anyone who contributed to the study but did not meet the authorship criteria. • Funding statement Please list the source of funding for each author. Because the schedule for publication is very tight, it is a condition of publication that you submit the revised version of your manuscript within 7 days (i.e. by the 10th July 2016). If you do not think you will be able to meet this date please let me know immediately. To revise your manuscript, log into https://mc.manuscriptcentral.com/rsos and enter your Author Centre, where you will find your manuscript title listed under "Manuscripts with Decisions". Under "Actions," click on "Create a Revision." You will be unable to make your



15

revisions on the originally submitted version of the manuscript. Instead, revise your manuscript and upload a new version through your Author Centre. When submitting your revised manuscript, you will be able to respond to the comments made by the referees and upload a file "Response to Referees" in "Section 6 - File Upload". You can use this to document any changes you make to the original manuscript. In order to expedite the processing of the revised manuscript, please be as specific as possible in your response to the referees. When uploading your revised files please make sure that you have: 1) A text file of the manuscript (tex, txt, rtf, docx or doc), references, tables (including captions) and figure captions. Do not upload a PDF as your "Main Document". 2) A separate electronic file of each figure (EPS or print-quality PDF preferred (either format should be produced directly from original creation package), or original software format) 3) Included a 100 word media summary of your paper when requested at submission. Please ensure you have entered correct contact details (email, institution and telephone) in your user account 4) Included the raw data to support the claims made in your paper. You can either include your data as electronic supplementary material or upload to a repository and include the relevant doi within your manuscript 5) Included your supplementary files in a format you are happy with (no line numbers, vancouver referencing, track changes removed etc) as these files will NOT be edited in production Once again, thank you for submitting your manuscript to Royal Society Open Science and I look forward to receiving your revision. If you have any questions at all, please do not hesitate to get in touch. Best wishes Andrew Dunn Senior Publishing Editor, Royal Society Open Sceince on behalf of Essi Viding Subject Editor, Royal Society Open Science [email protected] Associate Editor Comments to Author: Editor's note: Please address the remaining concern of reviewer 1 (listed here as 'Reviewer 2'). Comments to Author: Reviewer: 2 Comments to the Author(s) The extended experimental design section has clarified most of the issues previously raised. However, the references to group identification remain a concern specifically, that the identification differences between face-to-face and mediated communication in the context of ‘the group’ have not yet been addressed. Also, I note the frequent use of the qualifier ‘likely’ when describing group processes and that there is no discussion of the implication of the findings from the research for group processes. The context described for the submission is an open discussion forum involving large numbers of people and the collective emotions that ensue. Why is reference to ‘the group’ relevant? My preference would be for all reference to group identification to be omitted from the Introduction.



16

Reviewer: 3 Comments to the Author(s) The authors should be congratulated for the efforts they invested in meeting the reviewers' remarks and suggestions and in improving considerably the readability of their manuscript.

Author's Response to Decision Letter for (RSOS-160059.R1)

See Appendix C.



Dear editorial office of Royal Society Open Science,

We submit a new version of our original submission, taking into account the

comments by the reviewers:

- Both reviewers suggested to focus on the interactivity of the experimental

setup to differentiate it from previous communication studies. We are very

grateful for these suggestions, which motivated the most significant addition

in this resubmission. We analyzed how emotions trigger the expression of

positivity and negativity, as measured by the expression function in the agent

based model. Section 5c explains the methods and Section 3bii includes our new

results: Two different sentiment analysis techniques show that positive and

negative expression through text are strongly dependent on valence, while

generally independent on arousal, one of the assumptions of the agent-based

model of Schweitzer and Garcia (2010). We summarized this new important part

of the article in the abstract:

"When participating in an online discussion, the content of participants'

expression depends on their valence, and their arousal significantly

decreases afterwards as a regulation mechanism. We illustrate how these

results allow the design of agent-based models to reproduce and analyze

emotions in online communities. Our work empirically validates the

microdynamics of a model of online collective emotions, bridging online data

analysis with research in the laboratory."

- Reviewer 1 raised the question of the methods used for measuring emotions

based on Likert scales. We extended our explanation for the choice of Likert

scales in Section 5a:

"After reading the posts of a thread, participants provided subjective reports of their

emotions on three 7-point Likert scales to assess the subjective emotional response.

With view towards the requirements of the repeated measurement situation (10

threads), we aimed to obtain sufficiently reliable measures while the relatively weak

and fleeting emotional states that can be elicited by forum posts could still be

expected to be present. For this purpose, e.g., even the two relatively short 10-item

scales of the Positive and Negative Affect Schedule (PANAS, [63]) would have been

too long and not sufficiently focused on the immediate emotional response to an

individual emotional stimulus such as a forum thread. Instead, our measurement

situation was much more comparable to experimental designs assessing emotional

responses to emotional images, sounds, or words in the laboratory [64,65]. These types

of designs typically measure valence, arousal, and sometimes dominance (power)

via single-item graphical scales such as the Self-Assessment Manikin (SAM) [66].

Single-item Likert-type measures, if well phrased and designed, are not necessarily

worse than multiple-item scales [67]. For example the SAM has been the basis for

the validation of the extremely well cited International Affective Picture System

(IAPS) [68]. However, the graphical version of the SAM requires some additional

instructions and explanations that are, ideally, delivered by an experimenter who can

respond to questions. In addition, valence has recently been increasingly discussed as

a potentially two-dimensional construct allowing co-activation of both positive and

negative feeling states, such as in the case of bittersweet mixed emotions [69]."

And discussed the possible multidimensional extensions in the conclusion:

"The present research required very brief measures to avoid subject

fatigue and dissipation of the emotional impact of having read the short

threads, which limited the number of items that could be posed repeatedly.

Compared to single-item Likert-scales, a two-dimensional single-item measure

such as the Affect Grid [54] could have been even briefer, once sufficiently

explained. However, this type of measure still faces conceptual issues [55],

and specifically the good validity of the Affect Grid reported by the original

authors [54] has been found to be only moderate by subsequent research [56]."

Appendix A



- Following the comments by the reviewer 1, we added a detailed explanation

on how and why we compute valence from the participant responses:

"However, no systematic evidence for mixed-valence emotions was observed.

Cronbach's alpha=.88 across all positive and the inverse of all negative

valence judgments suggested sufficient overlap between both items to justify

integration of both scales into a single measure by subtracting the value of

negative affect from the value of positive affect and mapping the result to

the scale of [-1,1]."

- Reviewer 2 suggested a series of improvements to the introduction and the

explanation of emotional agents. Following those comments, we improved the

review of previous works and the framing of our approach in the introduction:

"Intergroup emotions theory, based on appraisal theory [1–3] and the social

identity perspective [4] suggests that dynamic variability of emotions over

time can be found not only at the individual level but also at the level of

group emotions [5]. More specifically, emotional experience has been found to

elicit the social sharing of emotion in the construction of emotional climates

[6,7], via processes that are likely to involve group identification [8],

emotion contagion [9], conformity with group norms [4], and social or group-

based regulation of emotion [5,10] to name only a few. It is likely that such

processes are not limited to face-to-face encounters, in particular since many

of the sources of over-time variability of group-based emotion may be at least

partially of an informational nature [5]. Participatory online communities,

such as social networking sites or discussion fora, provide a thriving medium

for the emergence of such collective emotional phenomena."

"Agent-based models can provide new insights into social psychology [8],

unifying different models of social interaction into a comprehensive

computational representation of emotions that can encompass a variety of

aspects, including identification during online social interaction [32,33],

changes in self- categorization as a group member [34], emotional contagion

[9,23], as well as affective- discoursive patterns that may help to reconnect

discourse studies with novel research on affect and emotion [35]. Towards this

aim, the modeling in the present paper will be based on the valence and

arousal circumplex of core affect [36–38]. This view is not specific to

emotion, but provides some grounding to the application of our techniques in

social psychology. Regarding emotional dynamics, some works within

mathematical psychology provide support for the usefulness of using agent-

based modeling in such contexts [39]."

- Reviewer 2 suggested that we show averaged dynamics on Figure 3. We did not

show an average on Figure 3 because our point is to illustrate individual

dynamics, not to show again the results of the regression methods. We

clarified this point by adding the following text to the explanation of the

figure:

"This illustrates the dynamics of an individual, rather than analyzing the

average response as shown in the regression results of Table 2."

- Following the comments by reviewer 2, we improved the explanation of the

ethical guidelines followed in our experiments:

"Conforming with standard ethical guidelines, participation in all three of

the above studies was entirely voluntary, and participants were informed that

they could quit participating in the study at any point in time without any

negative consequences. Participants provided informed consent at the beginning

of a study, and all participants were fully debriefed at the end of each of

the studies in this research. Furthermore, to avoid any negative impact on



participant's emotional well-being during or after taking part in the study,

the study design provided a balancing positive emotional stimulus for each

negative stimulus that was presented, thus preventing any buildup of negative

emotional states. In the laboratory studies, participants were furthermore

given the opportunity to ask questions to the experimenter who ensured that

there were no potentially lingering negative effects of the negative forum

threads at the end of the study."

We are very thankful to both reviewers and the editorial team for their time

and effort reviewing our submission. We commend this paper to you, and look

forward to hearing from you at your earliest convenience,

Yours Faithfully,

David Garcia, on behalf of all authors.



The dynamics of emotions under online interaction

Response to Referees

David Garcia, Arvid Kappas, Dennis Küster, Frank Schweitzer

We thank reviewers for their input. In the following, we explain

how we have assessed all the comments and suggestions of the reviewers:

> Reviewer: 2

>

> The authors have improved this submission significantly. However, a major

> problem remains; the assumption that the processes that result in shared

> emotions online are synonymous with those experienced face-to-face (as

> expressed at line 14 of the Introduction). There has been considerable

> debate and associated research on the relational implications of mediated

> communication and the findings that have relevance for the design of the

> Cyberemotions modelling framework should be addressed.

We thank the suggestion of the reviewer, who points at the necessity to

develop more the link between face-to-face emotional interaction and our case

of study. To clarify the similarities without imposing an equivalence between

processes, we comment in the new introduction:

"It is likely that such processes are not limited to face-to-face encounters,

in particular since many of the sources of over-time variability of group-based

emotions [5] and collective emotions [6] may be at least partially of an

informational nature."

> There is a related issue. The date of publication for reference 33, included

> as an example of online identification, is 1978 which predates the

> opportunity to communicate with others through forums and other social media

> by several years.

Our previous citation to that reference contained an unnecessary context of

online interaction. To clarify the issue the reviewer mentions, we rewrote the

mentioned sentence:

"... including identification processes [34] that can take place during online social interaction [35] ..."

> The report is focused on collective emotions. The new material in the

> Introduction focuses on group emotions. The psychology of group

> interpersonal relations does not necessarily translate to collective

> emotions.

In the current version of the article, we identify the line between collective

emotions and group emotions, citing seminal works on the relationship between

both. To make this point clear, we rewrote part of the second paragraph of the

introduction:

"Collective emotions are defined as states of an community in which a large

number of individuals share one or more emotional states [6,11]. While the

differences between face-to-face and online interaction are evident,

collective emotions can also emerge in online communities [12,13], a topic

that is receiving rising attention in the literature (see chapters 25 to 28 of

[13]). In the field, collective emotions can be found frequently in online

communities, including spontaneous fights in political fora [14], the

spreading of Internet memes [15] and political movements [16], or user-

generated content on YouTube that may, under certain circumstances, become

viral through the excitement of thousands of users [17,18]."

> The discussion topic as illustrated by one message (S1) appears to be

> emotive and is why the ethical criteria were requested. The report claims

Appendix B



> to be following the emotional valence and arousal of the interaction as it

> evolves. Indeed it is a significant advantage of the approach. However, the

> account of the ethics in Section 5a implies that the discussion thread was

> manipulated in some way – see lines 15,16 - that in fact participants did

> not interact with the discussion as it evolved.

Following the suggestion by both reviewers, we moved the methods section to a

more prominent part of the article and extended the "Experiment design"

section to include details about the manipulations, participant actions, and

demographics of participants. In addition, we explain the added the Ethics

Statement with more details about the guidelines followed to design the

experiments.

> Reviewer: 3

> The notion of "models of individual emotion dynamics" is introduced with a

> list of examples (p. 3, "cellular automata", "bifurcations in dynamical

> systems", "Brownian motion"). The reader might appreciate being explained

> the concept in some details.

We appreciate the comment by the reviewer, as our previous version of the

introduction did not illustrate well how agent-based modelling has been

applied to emotions. We added short descriptions of those previous modelling

approaches, to explain better the concept and the research background:

"For example, coherence in self-evaluation has been modelled as a cellular

automaton [42], in which memories are represented by cells and the attitude of

an individual to certain memory evolves depending on its relationship to

other memories, leading to the emergence of self-esteem. Concepts of dynamical

systems can also be applied to model emotions, for example explaining fight or

flee reactions as bifurcations [43] in which the emotional state of an

organism can sharply change depending on a control parameter. Furthermore, the

principle of Browinan agents [30] has been proved useful to analyze the

temporal evolution of core affect [33], modeling changes in emotions as

equations with a deterministic and a stochastic component."

> A bit further in the text comes the "Cyberemotions modeling framework". In

> the sentence that follows, it is said that this framework "allows the

> creation of models of user emotions etc." But besides this definition by the

> use, nothing is said of what "Cyerberemotions framework" means and

> represents, where it comes from and so forth

To give some background about the origins and applications of the Cyberemotions framework,

we added the following text to the introduction:

"As part of an interdisciplinary collaboration to understand the dynamics of

collective emotions in online communities [46], the Cyberemotions modeling

framework [12] was designed to provide generative mechanisms of online

collective emotions, explicitly avoiding the pitfall of using ad hoc

assumptions and implausible dynamics. This framework allows the creation of

models of user emotions under different kinds of online interaction, linking

collective behavior with individual dynamics in the presence of online

interaction mechanisms."

> The model that is at the core of the paper is described under the title 2

> "Emotional agents model". I had a hard time to go through it and to figure

> out exactly what the variables were. As far as possible, I would prefer

> reading first a detailed and clarified exposure of the variables in the

> model and their interrelations, followed in a later step by the set of

> equation formalizing them. As an illustration of the difficulties I met,

> "expression s" is first mentioned in the legend of Figure 1. It is only



> later on p. 4 that the notion of "expression" shows up in the text supposed

> to describe (articulate!) the model. And it is lost in a sentence stating

> that "the dynamics of the perceived emotional content (…) are influenced by

> the emotional content of the expression of the agents, si(t)". The notion of

> "expression rules" emerges a number of times in the article but as far as I

> have seen, it is never introduced or defined.

We thank the reviewer for these comments. We revised in detail the model

description, adding explicit references to all variables in the first

paragraph including the following text:

"Agents have an expression variable s_i(t) that quantifies their displayed

emotions through online posts, which are aggregated with the expression of

other agents in the online field of interaction h."

We follow the suggestion of the reviewer to verbally explain the dynamics

before the equations, adding the following explanation early in the section:

"Thirdly, the expression of an agent through s_i(t) is assumed to be triggered

by expression rules that determine the value of s_i(t) based on the emotional

state of the agent, leading to a feedback of expression that changes in the

emotional state of the agent after expression. "

> As regards the Methods, I regretted that the methods section took place only

> at the end of the paper. But this might be a specific requirement of the

> journal. In any case, it is impossible to go through the Results and

> Discussion sections without a minimal knowledge of what was done in study 1,

> 2 and 3. So, this minimal information could be sketched before the beginning

> of Section 3.

We followed the suggestion of the reviewer and moved the methods section

before the results section.

> The reader might want to know the context of the studies (what was said to

> participants?). Information available about studies 1 and 2 is really

> minimal, and virtually absent for study 3—which should certainly be

> corrected.

As also suggested by the other reviewer, we improved the explanation of all

experiment designs in this new revision.



We thank the editor and reviewers for their input. In the following, we explain

how we have assessed all the comments and suggestions of the reviewers:

> Reviewer: 2

>

> The extended experimental design section has clarified most of the issues

> previously raised.

>

> However, the references to group identification remain a concern

> specifically, that the identification differences between face-to-face and

> mediated communication in the context of ‘the group’ have not yet been

> addressed. Also, I note the frequent use of the qualifier ‘likely’ when

> describing group processes and that there is no discussion of the

> implication of the findings from the research for group processes. The

> context described for the submission is an open discussion forum involving

> large numbers of people and the collective emotions that ensue. Why is

> reference to ‘the group’ relevant? My preference would be for all reference

> to group identification to be omitted from the Introduction.

We rewrote the introduction to clarify the relation to group processes and not

to refer to group identification, as suggested by the reviewer.

Best Regards,

David Garcia, Arvid Kappas, Dennis Küster, and Frank Schweitzer

Appendix C



the dynamics of emotions in online interaction - open...

Documents