Assessing reactive violence using Immersive Virtual Reality.

Assessing levels of aggression-specifically reactive violence-has been a challenge in the past, since individuals might be reluctant to self-report aggressive tendencies. Furthermore, experimental studies often lack ecological validity. Immersive Virtual Reality (IVR) offers a reliable, ethically safe environment, and is the most realistic virtual simulation method currently available. It allows researchers to test participants' aggressive responses to realistic provocations from virtual humans. In the current study, 116 participants completed our IVR aggression task, in which they encountered avatars who would either approach them in a friendly or provocative fashion. Participants had the option either to shake hands or hit the virtual human, in congruent and incongruent trials. In congruent trials, the response required of the participant matched the approach with the avatar (e.g., hitting the avatar after provocation). In incongruent trials there was a mismatch between the avatars approach and the participants required response. Congruent trials were designed to measure the immediate reaction towards the virtual human, and incongruent trials to assess response inhibition. Additionally, participants also completed traditional questionnaire-based measures of aggression, as well as reporting their past violent behaviour. We found that the immediate aggressive responses in the IVR task correlated with the established questionnaire measures (convergent validity), and we found that the IVR task was a stronger predictor of past violent behaviour than traditional measures (discriminant validity). This suggests that IVR might be an effective way to assess aggressive behaviour in a more indirect, but realistic manner, than current questionnaire assessment.

Introduction

The Crime Survey for England and Wales recorded 1.2 million incidents of violence in 2020 [1]. There is thus an urgent need to further investigate and understand the causes of such violence, in order to come up with a means to reduce it. Aggression can be defined as any intentional behaviour to harm another person who does not want to be harmed [2].

Displays of aggression can be divided into impulsive, which is affective and reactive, and instrumental, which is proactive [2]. It has been suggested the impulsive subtype is related to poor emotion regulation and poor impulse control [3], which is especially present in situations in which individuals are provoked socially [4]. Work by Billen et al. [5] determined that crime and recidivism (the tendency of a convicted criminal to reoffend) were strongly associated with levels of self-control, which involved factors of coping skills as well as impulsivity. Indeed, Gottfredson and Hirschi [6] proposed self-regulation/self-control to be the strongest predictor of crime and aggressive acts. Furthermore, triggers for aggressive behaviours may involve discrete social dynamics [7] and in order to produce a none-aggressive response to another person, some individuals might activate inhibition and self-control to consciously inhibit their aggressive behavioural impulses.

Previously, observation and self-report have been the most common way to assess levels of aggression in individuals. Specifically, researchers have most often implemented measures of the variables that are thought to underlie and precipitate aggression such as irritability, hostility, impulsivity, and anger [8-10]. For instance, to assess facets of aggression, Buss and Perry [9] developed the Aggression Questionnaire (AQ) as a refined and expanded version of their earlier Buss-Durkee Hostility Inventory (1957). Probably the most widely used assessment, The State Trait Anger Expression Inventory-2 (STAXI-2) [11] uses manifested anger and an individual’s self-reported expression of it as a measurement of aggression. Using the STAXI-2 as a predictor for aggression is common, due its potential ability to predict some form of aggressive behaviours, such as institutional aggression [12]. However, studies investigating the relationship between the STAXI score and observed aggressive behaviour, have shown mixed results. For instance, Cornell, Peterson, and Richards [13] found that it failed to predict previous aggression in juvenile offenders, but that it was successful in predicting future aggression at the 3-month follow-up. Additionally, it was found that the STAXI subscale ‘state-anger’ was not correlated with aggressive behaviour [14]. In forensic settings, it has been found that the STAXI highly correlates with impression management (IM; i.e., giving answers which are perceived to be socially acceptable/desirable), whereby patients engaging in IM self-report significantly less levels of outward, and inward anger and higher levels of anger control [15]. Indeed, implementation of self-reported measures of aggression can lead to biased responses from participants due to the influence of social desirability [16]. Bech and Mak [17] evidenced an inverse relationship between measures of social desirability and hostility, whereby participants motivated by social approval reported less hostility than those who placed less value on social acceptance. Furthermore, Suris et al. [18] stated that utilizing self-report indicators of aggression often proves difficult as participants frequently share elements of higher order constructs and therefore are interrelated to the point of sharing common variance. Samples used in previous studies often implement contrastingly different assessments of the constructs of violence and aggression and therefore this might contribute to the low correlations within criterion measures. Recently, Berlin et al. [19] determined that self-reports of physical aggression where highly correlated with clinician’s assessment of aggression. However, both were not related to violent offence records.

Noting the potential limitations of self-report and introspective measures, researchers have developed studies of aggressive behavioural acts, involving tasks to induce frustration within the laboratory. Instruments include continuous performance tasks (CPT) such as the Integrated Visual and Auditory CPT [20] and interactive provocation tasks such as the Point Subtraction Aggression Paradigm Task [21] all suggested to potentially induce frustration. The Hot Sauce Paradigm [22] used measurements of hot sauce knowingly applied to another person’s food by a participant (where the consumer disliked the hot sauce) as a behavioural index of aggression in participants. Lieberman et al. [22] claimed this method overcomes limitations of other behavioural paradigms because participants perceive the potential for real harm to come to the target person. However, the extent to which hot sauce administration can be regarded as hostile physical aggression, and whether such laboratory studies hold ecological validity that is generalizable to a real-world setting, is debatable [23].

More recently, Verhoef et al., [24] developed the first pilot study, assessing children’s levels of aggression using virtual reality. In this study, 32 children aged 8–13 years were immersed in a virtual classroom in which they encountered several scenarios. For instance, another virtual child might push over a large toy tower as an act of provocation. The researchers found that the Immersive Virtual Reality (IVR) experience was greatly enjoyed by the children and that its behavioural observation results correlated strongly with the traditional paper pencil assessment of aggression. They also found that the IVR measure showed higher discriminant validity by being able to better determine individual differences in the level of aggression in children.

Virtual Reality was previously more commonly used in studies as an intervention tool to reduce aggression, with the hope that it could potentially be used as an intervention in clinical practice in the future. In a recent review, Dellazizzo et al., [25] examined previous studies, which have developed virtual reality interventions to reduce violence in incarcerated youth, in particular focussing on youth offenders who suffer from schizophrenia. They found few studies who have used IVR as an intervention tool for aggression but reported that preliminary studies using IVR showed some promising results, by reducing anger, as well as improving conflict resolution skills in teenagers. For example, Smeijers and Koole [4] developed a study protocol for an upcoming study, using an IVR game as an add-on tool to manage and reduce aggression, by using avoidance movement training to angry faces in forensic psychiatric outpatients. In IVR, virtual humans meet the patients in a virtual shopping street, approaching them in either a friendly or unfriendly manner. In the task the patients are then instructed to lean forward to agreeable avatars and lean backwards to disagreeable avatars. The authors speculate that this training now delivered in IVR will more significantly help to reduce anger and aggressive behaviour in the patients.

Klein Tuente et al., [26] applied virtual reality aggression prevention therapy to a sample of 28 offenders inpatients who displayed aggressive behaviour. The training involved 16 one-hour sessions in IVR, in which participants encountered adult virtual humans provoking them. The authors found that the participants’ self-reported or observed aggression was not reduced by the training. They did though find that the participants’ hostility and self-control significantly improved, although the authors reported that these changes were not sustained at the 3-month follow up.

Our study investigated self-control and response inhibition in association with aggressive behaviour, using a novel IVR task to assess levels of reactive aggression. The use of IVR in recreating simulated social scenarios has been demonstrated to elicit realistic responses by human participants; equivalent to real world responses [27]. Rovira, Swapp, Spanlang, and Slater [28] proposed IVR to be an effective measure in the study of violent situations, with high ecological validity. Furthermore, Rovira et al. [28] suggested that scenarios in IVR can elicit realistic responses via engagement of participants if the issue of plausibility is addressed correctly. Plausibility encompasses not only the credibility of the scenarios presented to participants, but also that their actions within the virtual world can have appropriate responses and are recognised within the illusionary virtual world. These issues of plausibility are addressed through sensorimotor contingency and domain design, creating a pragmatic method of measuring acts of aggression. The virtual human avatars used in our IVR task needed to act in a realistic manner (i.e., where possible exhibit random body movements, synchronizes animated moving lip movements when speaking, and make mild gestures that as closely as possible mimic the natural physical responses observed in real humans). Finally, similar to Smeijers and Koole [4] we also included elements of the ‘approach avoidance modification paradigm’ (i.e., forward movement as indicator of approach/attack and backwards movements as an avoidance/inhibition measure).

The overall aim of our study was to test our new IVR task as an assessment to measure levels of reactive aggression. We propose the hypothesis, that this new task would show convergent validity to traditional measures and that it would be a strong predictor of past violence. Specifically, we investigated if the response times to provocation by virtual humans (i.e., time to exhibit aggressive responses following the provocation) were related to the level of self-reported aggression. This suggests that the faster someone would exhibit the aggressive response after provocation, the more accessible this response option might be to that person, suggesting potential more impulsive aggressive tendencies. Furthermore, we examined if IVR response times were a better predictor of past violence than self-reported measures of aggression.

Method

Participants

Participants were recruited through the University’s online participation system, which provided a sample of participants who were mainly students, but included other non-student individuals who had voluntarily signed up to the university’s participants system. Out of the 116 participants (Mage = 25.91, SD = 12.43; 30 male) who took part in this study, 6% of the participants had attained education to a GCSE level, 72.4% to an A-level, 19.8% had a university undergraduate degree, and 1.7% had postgraduate education. Participants were reimbursed with either financially reward or with course credit. The study received ethical approval from Plymouth University ethics committee and we obtained written consent from every participant.

Materials

The State-Trait Anger Expression Inventory-2 (STAXI-2) [11]

The State-Trait Anger Expression Inventory (STAXI-2) included 57 items on six scales. The six scales were: state-anger, trait-anger, anger expression-out, anger expression-in, anger control- out and anger control-in. A sample item was “I control my urge to express my angry feelings.” For state anger, ratings were given on a 4-point scale (ranging from 1 = not at all to 4 = very much so). Trait anger measured the predisposition an individual possesses to become angry; the ratings were also given on a four-point scale (ranging from 1 = almost never to 4 = almost always). Anger expression was measured by the anger expression index, which evaluated the person’s propensity to express anger either outwardly to others or inwardly to themselves. Anger control indicated the level to which an individual prevents angry feelings by reducing them being expressed towards the other person or by reducing them internally. It was noted that the internal consistency of the subscales varied from α = .82 to α = .75 [11] (In this α = .72; α = .83 for state anger and α = .57 for trait anger; α = .54 for anger expression, α = .74 for anger control).

Reactive Proactive Aggression Questionnaire (RPAQ) [29]

This is a 23 item self-report questionnaire to assess reactive (following provocation) and proactive (instrumental) past aggression. Items are rated from 0 = never to 2 = often. A sample item for reactive anger includes “Became angry when others threatened you”. Cronbach’s alphas for the reactive and proactive scales were reported to be of 0.84 and 0.86, respectively.

The immersive virtual reality task

An Oculus Rift headset and a pair of Sennheiser headphones were used for the IVR task. The IVR program ran on a desktop PC and the participant was seated at a computer desk. There was a 22” LCD computer monitor located immediately in front of the participant so the experimenter could also monitor task progress. A joystick was used by participants to give the relevant responses and speed of response and the type of the responses were automatically recorded by the program. In the task individual avatars greeted the participant either pro-socially such as “Hello, how are you?” or non-pro-socially e.g. “What are you looking at?” on a hypothetical nightlife city street (See Fig 1). The task consisted of two different sets of 15 trials, the first set requiring congruent social reactions and the second set requiring incongruent social reactions. During the experiment, tasks were selected from the two lists at random. In the congruent trials condition, participants were required to push the joystick forward, to punch unfriendly avatars following the interaction, or pull the joystick backwards, to initiate a handshake with friendly avatars. In the incongruent trials condition, participants were required to react in the opposite way, thus an avatar interacting in a friendly manner would require a proceeding ‘punch’ response by participants (joystick moved forward) and an interaction with an unfriendly avatar would require a proceeding reaction from the participant in the form of a handshake (joystick moved backwards) (See Table 1).

Fig 1

Illustrations of the IVR task.

Panel A shows the street that the participant’s avatar walks down. Panel B & C Close up of the virtual agents that require a response from the participant. Panel D Participant generating an aggressive action from their avatar and hitting a victual agent.

Table 1

Blocks in the IVR task.

The table rows show the 4 experimental conditions. The condition column indicates the 4 possible combinations of anti-social and pro-social behaviour in congruent and incongruent scenarios. Note that we termed both responses to the incongruent trials as response inhibition due to the incongruent nature of the responses. The Avatar behaviour column shows anti-social avatar behaviour is aggressive, whereas pro-social behaviour is friendly. The Participant’s Instruction to Act column indicates how the participant is asked to behave for each experimental condition. The construct to assess column indicates the relevance of the response time in each condition.

Condition	Avatar Behaviour	Participant’s Instruction to Act	Construct assessed
Anti-Social 1 (Congruent)	Aggressive	Aggressive	Response time: Aggressive response following provocation
Pro-Social 1 (Congruent)	Friendly	Friendly	Response time: Friendly response following friendly approach
Anti-Social 2 (Incongruent)	Aggressive	Friendly	Response time: Response inhibition following provocation
Pro-Social 2 (Incongruent)	Friendly	Aggressive	Response time: Response inhibition following friendly approach

The program recorded the response action as well as the response time from when the avatar started speaking until the participant started to move the joystick. The avatars voices were all male pre-recorded and lip-synced to the virtual human to create a realistic impression. Apart than the lip sync, there were no expressions of anger or approach in the avatars face or posture. The participant was embodied into a virtual body with a virtual arm, which enacted the reaction made towards the avatar.

To enable the statistical analysis of the IVR data, a file was created with the response times recorded over the trials. For all trials in each of the four conditions in Table 1, the average response time was calculated. For instance, for the 15 trials in condition 1, the average response time for that condition was then created. All condition’s average response times were then used for the correlational analysis. We hypothesise that shorter reaction times in the Anti-Social 1 condition (e.g., faster responding to provocation) indicated greater level of aggression.

Procedure

After informed consent, participants first completed the IVR task in the IVR laboratory followed by the questionnaires. This task order was adopted to avoid participants focussing on the assessment of aggression during the IVR task. Participants were informed that the IVR might be distressing and that they could end the procedure at any time if they felt uncomfortable. No participant reported that they felt uncomfortable and no participant left.

Results

General response tendencies

Firstly, we investigated the general response patterns within each block, calculating differences in response time overall between the conditions (See Fig 2).

Fig 2

Response time differences during the different conditions in the IVR task.

Bar height indicates the mean response time and the error bars indicate +/− the standard error.

Fig 2 shows the mean response times in the different conditions, suggesting that participants were generally faster to respond to anti-social avatars. We conducted a repeated measures ANOVA with condition as the repeated measures factors and found it to be significant F(2.74) = 92.73, p < .01. Greenhouse Geisser corrected post-hoc tests revealed significant differences between all conditions (all ps < .01).

Overall, 63% of participants made no errors at all in the congruent condition and 53.4% in the incongruent condition. Only a relatively small proportion of 13.8% of participants made more than 1 error in the incongruent condition and 19.8% in the congruent condition. We found that errors were not normally distributed (D(116) = .30, p < .01). A Wilcock test revealed that participants made significantly more errors in the congruent compared to the incongruent condition (Z = -2.86, p = .004).

Correlations with individual differences measures

Table 2 illustrates the Pearson correlations between the IVR task and the traditional measures. Aggression in the anti-social 1 condition showed a significant negative correlation with the total anger expression index (AX) of the STAXI-2, as well as with the reactive past aggression self-report. Furthermore, the self-report STAXI-2 scales also correlated with past reactive aggression.

Table 2

Pearson r correlations between IVR task and aggression measures.

	Pro-Social 1	Pro-Social 2	Anti-Social 2	Stage Anger	Trait Anger	AX	RPAQ	Pro-active	Reactive
Anti-Social 1	.68**	.63**	.72**	-.08	-.15	25**	-.20*	-.10	-.22*
Pro-Social 1		.55**	.67**	.04	-.06	-.13	-.80	.01	-.15
Pro-Social 2			.79**	-.01	-.01	-.09	-.14	-.12	-.18
Anti-Social 2				-.02	-.07	-.17	-.13	-.06	-.17
State Anger					.31**	.15	.34**	.16	.09
Trait Anger						.71**	.62**	.06	.26**
AX							.54**	.01	.14
RPAQ								.35**	.47**
Proactive									.52**

** is significant at .01 level;

* is significant at .05 level

We used Spearman’s correlation to determine none-parametric correlations with aggression self-report measures and errors made in the IVR task, however, we did not find any significant relationships (all ps > .05).

Finally, using backward regression, the final remaining model, including Trait Anger and IVR Anti-Social 1 (congruent condition) could significantly predict reactive past aggression. Model R Square = .104; p = .002 (See Table 3).

Table 3

Factors in the regression model.

Concept	B	T
Constant	6.39	2.89*
Trait anger	.20	2.6**
IVR = Anti-Social 1 (congruent 1 condition)	-.94	-2.1*

Discussion

The aim of this study was to investigate a novel, interactive IVR task to assess levels of reactive aggression after provocation by virtual humans. In particular, we compared responses in the IVR task to traditional self-report-based measures of aggression and its relationship to aggressive past behaviour. We found that the response time towards virtually hitting an avatar after provocation correlated with the overall anger expression index of the STAXI-2, as well as reactive aggression from the proactive-reactive past aggression questionnaire. Specifically, as predicted by our hypothesis, we found that only the response times to act aggressive following provocation (i.e., not the response times in all conditions) were related to self-reported levels of aggression. Thus, higher levels of self-reported aggression were correlated to lower reaction times on the IVR paradigm. Furthermore, we found that this IVR task component was a better predictor of past violent reactive self-reported aggression than the self-report measures.

These findings support previous IVR measures, who also started to find associations between IVR aggressive responses and traditional self-report measures [24]. Verhoef et al., [24] developed the first IVR pilot assessment for children and also found good convergent and discriminant validity for the task. Indeed, establishing IVR as a valid measure of aggression has important implications for future uses of this advancing technology. Rovira et al. [28] concluded that IVR may also be useful for rehabilitation of victims of aggression who might become disturbed by their behavioural response in a real-world situation. With further research, this conclusion could be expanded to include aggressors themselves who might benefit from the realisation of not only their actual behaviour, but also the consequences to the recipients and victims of their aggressive behaviour. Identifying risk factors and protective factors of aggression could allow for appropriate strategies and frameworks of intervention to be implemented more effectively, potentially yielding more successful outcomes for individuals, rather than interventions occurring reactively to violent or other types of aggressive behaviour.

IVR delivers an opportunity to measure immediate emotional responses [28], and in our task we suggest that the response time to display the aggressive response after provocation can be used an indicator of an individual’s accessibility of the aggressive emotional impulses in the ‘heat of the moment’.

Participants however, made also errors in the IVR task, which constituted making the opposite response to the one required (e.g., they were hitting the virtual human when a handshake was required). Errors in response will clearly add noise to the results, but should not affect the underlying trends, especially when their incidence is low, as in our study. To reduce the effect of such errors, future studies could investigate disabling the option to perform the opposite response on the IVR response device during the task, thereby eliminating the opportunity to make an inappropriate response.

Overall, a few participants (13.8% in incongruent and 19.8% in the congruent) made more than one error, and the number of errors were not correlated to the self-report measures. We found that people made more errors in the congruent condition compared to the incongruent one. This might indicate that the congruent condition was processed by participants in a more emotional manner (i.e., with an intuitive and impulsive response after the virtual human’s approach). In contrast, they might have engaged a more cognitive approach in the incongruent condition, in which they were concentrating more about providing the correct response rather than acting on impulse.

Limitations

Our current study involved mainly students, of similar age and educational background, with the majority of them being female. Physical violence has however been reported to be greater in men [2], thus future studies would need to include more male participants to further validate the IVR task.

We used our current sample of participants as the first attempt to validate our new IVR measure. Future studies are clearly needed to assess and replicate the findings using different samples of participants. Thus future research could also involve the IVR task being performed on a sample of clinical interest, such as violent offenders. Previous research has proposed and demonstrated that behavioural measures perform better in populations where the targeted variable occurs at a higher rate of incidence and magnitude [30]. Direct comparisons of sample characteristic would allow for further analysis of the relationship and constructs of aggression and executive functioning, where clinically relevant features such as psychopathy or violent tendencies may moderate the relationship between self-control and aggression. Such comparisons give potential for an accepted taxonomy of anger and aggression and resolve the issues raised by Lee and DiGiuseppe [31] by better informing targets of treatment and rehabilitation.

A further limitation of the current study is the nature of self-report measure for past reactive aggression, as social desirability potentially confounds the self-reported measurements of aggression and anger. Future studies might implement observational methods in forensic settings as a variable for discriminant validity. Furthermore, in our IVR task, the responses were made using a joystick. In order to increase ecological validity of the task even more, other devices (i.e., also those simulating force) could be used to make the hitting/shaking hands movement not only visually, but also physically more realistic. Indeed, in a recent study using a morality IVR task, we implemented a robotic maniulandum to more realistically produce measures of force and physical touch into the task [32]. Additionally, order effects of the IVR task and presenting the questionnaire could be investigated in future studies.

We suggest our current study provides a framework and data for future research using the IVR Aggression Task. The future possibilities and benefits of IVR as an effective tool in psychological research are promising. Additionally, IVR is usually regarded as an engaging and ‘fun’ tool, thus making it potentially more attractive for participants to engage with the technology. In addition, IVR technology affords ecological validity without compromising experimental control, thereby increasing experimental impact, and reducing replication difficulties. Furthermore, since the tasks investigates aggressive responses to virtual humans, it does not involve aggressive responses to real humans, making it more ethically safe and acceptable. The advancement in IVR Technology brings great potential for developments in psychological experimentation, theory, and interventions. Thus, since it has been suggested that reducing reactive aggression is of top priority in forensic settings [26], effectively measuring it might be a first step to help to develop evidence-based interventions to reduce it. We thus believe this new IVR task a first step in its application as an assessment tool for levels of aggression in forensic populations and to also to indirectly measure potential improvements after aggression treatment. This could be achieved by examining differences in our IVR task scores before, during and after an intervention. Future studies might also investigate the use of IVR at additional time intervals before during, after treatment. Additional measurements could also be added, combining the IVR assessment with direct physiological measures, such as heart rate and skin conductance.

Finally, we note that a presentation of some aspects of this work is available in an online YouTube video here: https://www.youtube.com/watch?v=xqPJKtFI42U

This was recorded at a TEDx talk in 2018 with the title: “The light and dark of emotional intelligence”.

3 Dec 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Stefano Triberti, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

3. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide

4. Please include your full ethics statement in the ‘Methods’ section of your manuscript file. In your statement, please include the full name of the IRB or ethics committee who approved or waived your study, as well as whether or not you obtained informed written or verbal consent. If consent was waived for your study, please include this information in your statement as well.

5. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Partly

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This manuscript described a novel IVR task, designed to assess aggression, and preliminary findings on its validity. Although I do concur with the authors that IVR provides completely new possibilities to assess (and treat) aggression, the manuscript requires substantial revisions before publication can be considered. I would love to see a revised version. Please see detailed suggestions and comments below.

Introduction

In general, the introduction needs to be revised and the authors must make sure to increase clarity in conceptualizations in relation to aggression. As it stands now, the authors shortly mention key terms that are used very broad but would require more context. Please make sure to be specific and not overgeneralize, since this may lead to crucial misinterpretations. There are several, recent references which could be used to improve the introduction. For instance, see Billen et al., 2019; Berlin et al., 2021; Martens et al., 2019.

p. 3, paragraph 1:

* Please provide the context for this statement, since it otherwise is too broad: "The prevalence of aggressive and violent acts is significant and requires attention and investigation."

* What do the authors mean by "Often dichotomous, violent acts arising from

aggression may be predatory, impulsive, reactive, or defensive and such behaviours may arise

from situational or environmental factors.". Please clarify what is meant by "dichotomous" in this respect.

p. 4, paragraph 1:

* This sentence comes without context, please remove or provide context: "Our novel IVR aggression task allows us to test levels of response inhibition in complex virtual social environments."

p. 4, paragraph 2:

* The AQ does not specifically target only hostility, the current phrasing can be misleading and I this recommend revision.

p. 5, paragraph 2:

* CPT tests are not conceptualized as aggression measures, which is how it may be read as is written currently. It may, however, induce frustration. Please remove or revise.

p.6, paragraph 2:

* VR is not a common intervention tool in aggression management, but is increasingly being tested as such in different forensic settings.

* As is written now, the reference to Klein Tuente (please revise to Klein Tuente) et al seems to refer to a similar intervention as Smeijers and Koole (VR-GAIME; which is only used as an add-on in ART trainings). These are two completely different methods. I recommend to either provide details for both, or just present shortly and as two independent methods.

p.7, paragraph 2: I strongly recommend the authors to not suggest that they will test causality, since this would require a completely different research design than what is provided in the manuscript at hand.

p. 7-8: I recommend the authors to, clearly, write the scientific questions that are being addressed within the current study, so that this part can be compared to how the results are presented.

Methods

p. 8: please provide more information on the participants.

p. 8: please provide internal consistency values for the STAXI-2 scales in the current study (or have I misinterpreted the values as coming from the reference?).

p. 8: I would suggest the authors to use "instrumental" instead of "forward" when referring to proactive aggression.

Please provide a part describing the statistical considerations and analyses.

Results

p. 15: what measure in the IVR task was used for the correlational analyses? As it stands now, the reader has no possibility to interpret the values provided by the table. This interpretation could be facilitated by a part in statistical analyses in the methods section. This information is provided in the discussion, but must be clear already in the methods and results sections.

p. 16: the headline for Table 2 should refer to Table 3, and be followed by Table 3, I guess?

Discussion

My main concern on this part is that the authors use response time on the IVR task as a proxy for aggression in IVR. Unfortunately, I do not see any solid evidence-base for this assumption anywhere in the manuscript. I strongly recommend the authors to revise the aims of the study (p. 7-8) to better correspond to what was actually tested in the study, and rewrite the discussion with this in mind.

Minor comments:

* please proof-read once more, some words are missing in some sentences, misspellings occur (e.g. anti-social, pro-social, some references), and abbreviations should fist be introduce and then used consequently throughout the manuscript.

Reviewer #2: First of all, I want to congratulate the authors on their study. This manuscript concerns an important issue for society, using innovative and promising techniques. The possibility of finding more ecologically valid instruments is very interesting, for both empirical research and clinical practice. That is why I enjoyed reading the manuscript.

I have some major and minor issues with the manuscript in its present form, however, which in my opinion need to be addressed before it is suitable for publication. These are listed below.

1. In the abstract, it is not clear enough what the function is of the congruent and incongruent trials. Could this be explained in a few words?

2. In the Introduction, concerning the following sentence: “Often dichotomous, violent acts arising from aggression may be predatory, impulsive, reactive, or defensive and such behaviours may arise from situational or environmental factors.” For readers who are not familiar with aggression concepts it may not be clear what ‘dichotomous’ refers to, especially since multiple terms from different models/authors are subsequently mentioned randomly. Please explain the dichotomous character. Furthermore, be aware that terms of different authors are mixed together in this sentence. Maybe it is preferable to stick to one model (for example: proactive versus reactive aggression according tot Dodge, or instrumental versus impulsive aggression). And, finally, add a reference here.

3. In the introduction, the general textual structure is difficult to follow and it is too lengthy. For example, in the first part of the introduction, the authors shift from prevalence to dichotomous aggression types, to models (with in between again dichotomous aggression types), to (yet again) similar aggression subtypes, followed by – out of the blue- an IVR aggression task, which is only explained at the end of the introduction. Please look into the flow of your introduction to make it more easy to read (for example: choose to use only one set of terms to describe the dichotomy between reactive and proactive aggression, and do this in one paragraph instead of spread around the entire introduction). Furthermore, shorten the introduction considerably (for example the second paragraph (GAM / SIP) could be removed entirely without compromising the key message of the introduction, and the same goes for various other parts of the introduction (the AQ/STAXI descriptions and provocation paradigms could be discussed in less detail. Also, the intervention study using VR could be removed from the introduction, since the present study is not an intervention study (therefore not relevant in this section and, thus, less distracting for the reader), but it may be interesting to mention this study in the discussion as another field of application for VR in relation to aggression).

4. In the introduction: please first write the immersive virtual reality (IVR) and consequently in abbreviated form later (it is first mentioned in abbreviated form only in the first section at page 4)

5. Introduction, page 7, concerning the following sentence: These issues of plausibility are addressed through sensorimotor contingency and domain design, creating a pragmatic method of measuring physical acts of aggression. Could the authors explain how this is done?

6. In the final section of the introduction, be consequent in using past or present tense in describing the present study. And please do not jump to another previous study here (this closing part of the introduction should only describe the present study)

7. Introduction, last paragraph: is the aim of the study actually to develop an assessment? Should this not be: to determine if the (already) developed IVR-assessment is a good (valid) instrument to assess aggression?

8. Method: is it possible for the authors to give more information about the participants? Were these all students? Is it known whether they had problems with aggression in the past? And, if not, is this a study limitation? Should the task not better be assessed in clinical forensic groups who are known for high levels of aggression and where it is most likely to be used in the future? (this latter point is already appropriately discussed by the authors in the discussion section, however)

9. Method, STAXI: please be consequent in present or past tense. Can the anger control scales be explained briefly like the anger expression scales?

10. Method, the IVR task: can the authors give information of the degree of emotional expression on the avatar’s face and posture (in other words: hoe realistic is for example an aggressive or a friendly approach with respect to nonverbal features, apart from the lip synchronisation?)

11. Method: the authors describe that the participant sees a virtual arm to enact the reaction. However, the participants actually use a joystick to initiate this virtual response, which means that their physical sensorimotor perception is not congruent with the perceived visual response. Can the authors elaborate on what in their opinion this might mean for the ecological validity of the task (in the discussion section)?

12. Method, IVR: how is the response action operationalized? Please explain. For the results section: in the correlation and regression analysis it is not clear at all what type of variable is used here in the analyses to represent the IVR-conditions! This must be explained (in the discussion is mentioned that this concerns response time, but this needs to be clear in the previous sections as well). And it needs to be explained why the authors think that this measure is representative of actual aggression. Furthermore, nowhere in this article is described what the direction of the correlation is; only that it is significant. Thus, are shorter reaction times related to higher levels of aggression or are longer reaction times related to higher levels of aggression? And how do the authors explain this finding? (please elaborate on this in the discussion section) What is the relevance of it and is this known from literature?

13. Method, IVR task: can the authors explain what the function is of the congruent versus non-congruent conditions?

14. Procedure: did the authors investigate if the order of the presented blocks in the IVR task was of influence on the outcomes of the aggression questionnaires (in other words: was there a priming effect of the IVR- conditions on the questionnaires – for example when the last condition elicited an aggressive state preceding the questionnaires - and is there corrected for such an effect?)

15. Results, Table 2: What correlations are reported here? Pearson’s r? The RPAQ variables in the first column do not correspond with the RPAQ variables in the top row. (and please add RPAQ to the text above this table where correlations with reactive past aggression are described)

16. Results: last paragraph: please add that Anti-Social 1 was the congruent condition

17. Discussion: can the authors give an indication of the expected field of application for this tool? Do they see it as an instrument for empirical studies only or as an actual diagnostic tool in clinical practice? For example to assess the type and severity of a patient’s aggression problem, or to determine an individual’s improvement during aggression management treatment?

18. Discussion: A strength of the VR environment can be that it is fun to do, which might make individuals more willing to cooperate with such assessments, which is often a problem in forensic populations. The authors might stress this more.

19. Discussion: Might it be a valuable addition for the future to combine the IVR task with biomarkers such as heart rate or skin conductance to even better assess aggressive responses?

20. In the abstract is suggested that IVR is ethically safe. This point is not mentioned in the manuscript. Could it be explained in the discussion section how the authors see this?

21. References: 31, Klain Tuente is spelled incorrect (this should be Klein Tuente)

Reviewer #3: This is an interesting study showing the potential usefulness of IVR for assessment purposes and (potentially) for treatment of aggressive people. The main problem has to do with the sample composition: the authors should move the section 'Participants' to the start of the Results section. Moreover they should give many additional details about the participants: education; occupation; any history of mental disorders or psychological treatment; any history of offenses, etc. Once they have added these variables (hoping that they have this info!!), they should make additional analyses to see whether any of these important sociodemographic features is associated to different IVR responses. Incidentally, the sample is clearly composed by university students, as shown by the young mean age. Moreover, the large majority of the participants are women (30 males out of 116 participants): the authors must comment on this very important sample feature, since a century of research on aggression has shown that aggressive behaviour, of any kind, is largely more frequent and severe in males compared to females. They should have considered this during recruitment and should have tried at least to achieve a fair gender balance in the sample composition. Therefore the over-representation of young females represents a major limitation of this study and has to be acknowledged.

The potential utility of IVR for treatment purposes, including its utilization in clinical populations, has to be demonstrated and so far there is nothing which can support this claim: this has to be made clear.

There are a few typos which have to be corrected:

- page 3: para 2, add 'how' after 'describes'.

- page 5: line 10, drop 'this'.

- page 6: correct 'aggreagable' in 'agreeable'.

- same page, end page: add ' to 'participants'.

- page 7: line 2, 'change'

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Niki Kuin

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

31 Jan 2022

Please see the attached responses to reviewer letter.

Submitted filename: Response_to_reviewers_31.01.2022.docx

Click here for additional data file.

16 Feb 2022

12 Apr 2022

Please see the attached "responses_to_reviewers" letter

Submitted filename: responses_to_reviewers_30.03.22.docx

Click here for additional data file.

25 Apr 2022

Assessing reactive violence using Immersive Virtual Reality

PONE-D-21-27929R2

Dear Dr. Terbeck,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Stefano Triberti, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

27 Apr 2022

PONE-D-21-27929R2

Assessing reactive violence using Immersive Virtual Reality

Dear Dr. Terbeck:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Stefano Triberti

Academic Editor

PLOS ONE