Virtual Reality Training for Aggression Control in a Dutch prison-based population - a randomized controlled trial.

Violent crimes (violence directed against a person or matters) are still one of
the most common crimes in a Dutch prison-based population. Violent crimes alone
take up to 29% of all offenses. If property crimes with violence and sexual
assaults are also included, this number can increase up to half of all
offenses. When looking at the recidivism rates of these types of crimes, it has
been estimated that 22% of these offenders will return to prison within two
years after they have been released (*DJI in getal 2013-2017,* 2018). Research
also shows that individuals with violent crimes during their incarceration,
recidivate more often and even sooner than inmates without violent crimes
(Mooney & Daffern, 2015), which indicates the importance of treating
individuals with aggression problems in detention.

Although the terms *aggression* and *violence* are used interchangeably, it is
important to differentiate between the two terms (McGuire, 2008; Ramírez &
Andreu, 2006). The term *violence* is generally described as personal
(physical) violence committed by one individual to another and is mostly
defined and used in formal law (McGuire, 2008). Aggression is defined as *any
behavior to harm another person* (Ramírez & Andreu, 2006), which means it can
consist of different dimensions. Literature, in general distinguishes two main
types of aggression: reactive and proactive aggression (Crick & Dodge, 1996).
Reactive or expressive aggression is described as an impulsive, angry, hostile,
or defensive response to a frustrating or provocative situation. The goal is to
decrease an unpleasant internal feeling (for example the physiological arousal
or tension). Whereas proactive aggression is deliberate controlled behavior to
obtain the desired goal (Crick & Dodge, 1996; McGuire, 2008).
One important determinant of aggressive behavior is the hostile attributional
bias (HAB), which states that negative experiences in childhood and adolescence
may cause individuals to interpret behavior as more hostile, especially when
behavior is more ambiguous or unpredictable (Dodge et al., 2015). It is also
thought that these individuals are more likely to evaluate behavior as
aggressive and they expect more favorable outcomes for aggressive behavior
(Coccaro, Fanning, & Lee, 2017).
A more general and extended explanation of the hostile attributional bias is
the social information processing theory (SIP) (Dodge & Crick, 2007). Although
this model was initially developed to explain aggression in children, more
evidence is rising that HAB is also applicable in adults (Klein Tuente,
Bogaerts, & Veling, 2019; Lim, Day, & Casey, 2011). The SIP describes six steps
in which an individual processes information to interpret different situational
cues to choose appropriate behavior (Coccaro et al., 2017).
The first step includes encoding a situational cue (cognitive and emotional
cues). The second step includes the interpretation of the social cue, and
attributions are made. The third step is to consider the goals in the given
situation. In the fourth step, the potential responses are considered. The
fifth step consists of selecting a response. The last step is choosing a
response and act on it (Coccaro et al., 2017). Later, emotional factors were
also included in this model (changing the name to Social-Emotional Information
Processing (SEIP) model), including the role of negative affect in aggressive
behavior, which made the model applicable to both types of aggression (Coccaro
et al., 2017).

Many forms of aggression-based therapies have been developed, mostly consisting
of cognitive or behavioral methods, or a combination of the two (Shelton,
Sampl, Kesten, Zhang, & Trestman, 2009). The focus is mostly on reshaping
cognitions, improve problem-solving, exposure, and skill training (Shelton et
al., 2009). Research has shown that treatment for aggression in general works
in reducing recidivism rates (McGuire, 2008). In prison-based samples, highly
structured treatments with a focus on criminogenic factors showed to be more
effective. Also, a more therapeutic approach in an institutional regime seems
to be more effective (Auty, Cope, & Liebling, 2017). However, results in
prison-based populations are still inconclusive, mostly due to the low quality
of the studies (Auty et al., 2017; McGuire, 2008). McGuire (2008) indicated
that this can be best interpreted as: *absence of evidence rather than evidence
of absence* (p. 2591), stating that more and better quality research is needed.

For therapy to be successful, a couple of factors are of importance. One of
these factors is the motivation or the willingness of the offenders to change
their behavior (Jochems et al., 2012; McGuire, 2008; Smeijers, Bulten,
Buitelaar, & Verkes, 2018). Individuals may be convinced that they have no
problem or may have followed therapy earlier without success, demotivating them
to follow therapy again. In addition, most therapies are very theoretical,
which may not be successful for some specific populations. It is recommended,
for example for individuals with intellectual disabilities, to use less
language-dependent therapies (Simpson, Mizen, & Cooper, 2016). Lastly, a known
difficulty in imprisoned individuals is the inability to practice learned
behavior in real-life situations (McGuire, 2008).

A solution to the above-encountered problems may lay in the use of Virtual
Reality (VR). VR uses artificial computer-generated environments to imitate
real-world situations. Individuals wear a headset with controllers to move
around in the VR environment (El Beheiry et al., 2019). VR makes it possible to
practice situations in an interactive computer-generated environment and
therefore combines theoretical information with practical learning situations.
Situations can be altered as much as needed to fit the precise conditions in
which the individual wants to practice the problematic behavior (Freeman et
al., 2017). This technique may also trigger the motivation of individuals, as
it is new and interesting, and may also be attractive to the younger generation
since they grew up with technology and innovation.

The primary objective of this randomized controlled trial is to investigate the
effect of the VR-TRAC on aggression in a prison-based population.

This study will be a randomized controlled trial.

Procedure
1. All detainees meeting the inclusion criteria will be made aware of the study
through the psychologist, the case manager or the mentor on the ward. Also,
there will be flyers on the ward and information on a special prison tv
channel.
2. When a detainee is interested to participate, a research assistant will
visit him to give information about the study and check if they meet the
inclusion criteria. As screening with the Aggression Questionnaire (AQ) is
needed for assessment of eligibility, informed consent for the study is
obtained before the AQ is administered.
3. When participants are included in the study they are randomized to one of
the two conditions: VR-TRAC or waiting list.
4. Both groups will start with filling in the SAM weekly and the OSAB will be
filled in with staff members, also weekly (this will start four weeks before
treatment and will continue until four weeks after the last session has taken
place).
5. After four weeks, the other self-report questionnaires and vignettes are
filled in by the participants in both conditions for baseline measurements
(DERS NAS-PI, RPQ, BIS-11, ACE, and MATE). The research assistant will assist
with filling in the questionnaires (explain the content of the questionnaires,
assist when participants don*t understand the questions and check if any
questions are accidentally forgotten). To motivate the waiting list condition,
participants are given the option to play a game in the VR surrounding for a
maximum duration of half an hour. The surrounding is different from the
surrounding in which participants follow the VR-TRAC and does not contain any
aggression like situations.
6. Next, participants in both conditions will be scheduled for the role-play
assessment.
7. Participants randomized to the VR-TRAC condition will then start treatment.
Treatment consists of 16 twice-weekly individual sessions with a maximum
duration of 60 minutes per session.
8. After every session, the SRS is filled in and additional evaluating
questions are asked.
9. At the end of treatment, vignettes and baseline measures (only AQ, BIS-11,
DERS, NAS-PI, RPQ) are repeated in all participants. The research assistant
will assist the participants with filling in the questionnaires and
participants in the waiting list condition are given the option to play a game
in the VR surrounding.
10. All participants will be scheduled for the post-treatment role-play
assessment.
11. Two months after the end of the VR-TRAC sessions, self-report measures are
repeated. Participants in the waiting list condition are also given the option
to play a game in the VR surrounding.

The protocol that will be used in this study is based on the protocol that has
been developed by Klein Tuente et al. (2018). In our pilot (METc 2019/381,
registered under file number NL71610.042.19) we tested this protocol in a
prison-based population. After the pilot, some minor adjustments were made.
The current protocol is still based on the Social Information-Processing (SIP)
model of Crick and Dodge (1994) and consists of 16 sessions in total, with a
duration of 60 minutes each.
The first session will focus on the introduction of the training and
formulating goals for the participant. It is also important for the participant
to get acquainted with the VR (how it all works, but also how the VR world
looks like). Session two and three will focus on the early stages of
information processing (what is happening and what does it mean). Session five
through eight focuses on the late information processing stages (what goals am
I trying to achieve, what options do I have to react, what am I going to do,
and what is the reaction or behavior). Session 10 through 15 combines the early
and late stages, as all new learned behavior will be incorporated in the
interactive scenarios. To train the aforementioned stages, different
aggressive-inducing situations are formed in VR.

During the VR-TRAC sessions, patients wear headsets with controllers and walk
in a simulated virtual environment. The virtual environment is adapted to the
specific needs of the patients, with different themes (for example a store, bar
or mall) and avatars (for example a security guard, a group of females or males
with different ethnic backgrounds) to choose from. The trainer takes the role
of the avatar by using a microphone with voice distortion, controlling the
facial expressions and bodily movements throughout the session.
In the sessions, patients are able to train de-escalating behaviour in
interaction with the avatars. The VR-TRAC is a 16 twice-weekly individual
training session with a maximum duration of 60 minutes.

participants will undergo two types of measurements during the study
(self-report measurements and performance-based assessments). The self-report
measurements will approximately take an hour on each timepoint to fill in. The
vignettes will take approximately half an hour to fill in (before the treatment
starts and after treatment has ended). The weekly questionnaire on angry
feelings will approximately take a maximum of 10 minutes to fill in. Research
assistants will be available to support when questionnaires are filled in.
The performance-based assessments, which will take place before the treatment
starts and after the treatment has ended will take approximately half an hour.
Risks associated with participating with this study are cybersickness and
aggression during the sessions. The risks are kept to a minimum as the VR
scenarios to provoke angry feelings are build up slowly. Cybersickness is a
mild side-effect as a result of exposure to the VR environment and symptoms
reduce when VR is stopped. The risks of being on the waiting list are limited
because aggression therapy is not standard part of the therapy program in
prison. Detainees can register voluntarily for individual psychological therapy
(where they can also follow aggression therapy), but for following individual
therapy there is a waiting list as well. When they have participated in the
study they are offered the opportunity to follow the VR-TRAC after the last
follow-up.
We understand that filling in these questionnaires and participating in the
role-plays may be intensive, but we expect participants to benefit from the
training as they will learn to control their emotions, gain more insight in
their aggressive behavior and use de-escalating skills. We also think that
adding the role-plays and vignettes as an extra measurement will help to
understand the learned skills and if they extend to the *real-world*, instead
of focusing solely on the broad terms of aggression.