By: Lily Fitzpatrick
Inside the Trolley Problem–a VR-based research study that immerses participants in one of moral philosophy’s most famous hypothetical scenarios–I stood in a control room, surrounded by a dark, dismal industrial landscape. Between that and the robotic computer commands, flashing emergency lights, and alarming signals that a train was coming, I found myself a little tense (even with the volume set so low I could barely hear anything over the ambient noise of Studio X). When I had to stop the train by pushing a barrel of water to cause a temporary electrical short, I felt a genuine sense of time pressure and anxiety to save the five service workers on the track–even though they were animated. When my instructor eventually told the class to take off our headsets, I was in the middle of another life-or-death task–diverting the train to another track to save five different workers–and felt an initial impulse of reluctance to abandon it, despite my full awareness that it wasn’t real. I didn’t get far enough to face any moral dilemmas; early on in the study, the only choice is to save five people or sit back and let them die. But based on my experience thus far, I suspect that any future decision I made in the simulation would have seemed of the utmost importance (i.e. life-or-death), when it was really inconsequential. If I had, say, pushed one virtual person onto the tracks to save five others, I would have felt real internal conflict and real guilt, as if the lives of these nonconscious avatars truly mattered.
VR simulations, due to their immersive properties, have the ability to predict (within a reasonable degree of certainty) people’s genuine responses to given scenarios–an ability which will reach its full potential with continued technological development. Therefore, simulations may be an especially effective tool to train police officers in de-escalation tactics; experiencing virtual representations of real-life situations they will face, trainees can observe and receive feedback on their behavioral responses in a safe environment before confronting (and potentially harming) real people. Given the persistence of police brutality in the U.S., more departments should consider strengthening their de-escalation training programs with VR.
VR and the Prediction of Human Behavior
The potential benefits of VR in de-escalation training stem from the flexibility and immersive nature of simulations. As recognized in a paper by researchers Mel Slater and Maria Sanchez-Vives, titled “Enhancing Our Lives with Immersive Virtual Reality,” VR is a “powerful tool” that can be beneficially applied to the study of social and moral behavior. Besides simulating “extreme situations that are ethically and practically impossible in reality”–the Trolley Problem being a prime example–we can theoretically program virtual representations of any “real-life” situation imaginable (Slater and Sanchez-Vives). Therefore, we can simulate various types of police interactions with civilians, which officers-in-training can experience and respond to.
In order for such simulations to serve as a useful training exercise, the trainee must truly feel immersed in it. In other words, the program must successfully induce “presence,” defined by Slater and Sanchez-Vives as a combination of place illusion (the feeling of truly “being there” in the virtual world) and plausibility (the “illusion that the events are really happening”). The extent to which participants’ behavior in VR models how they would behave in real life relies on presence (Slater and Sanchez-Vives)–and conceivably, as VR technology continues to develop, its presence-inducing capabilities will become increasingly advanced, causing users’ responses to become increasingly authentic. Perhaps our simulations could eventually be as sophisticated as those represented in Veronica Roth’s Divergent series, with non-animated environments that the user’s senses cannot distinguish from real ones–but would we really feel emotions just as intensely as we would in real life? Would we behave exactly as we would in reality? Even if, in a refined version of the Trolley Problem simulation, I had the impression of touching the trolley controls with my own fingers (as opposed to glowing blue ones, devoid of tactile sensation), and even if the people on the tracks had not been animated, I would still retain the knowledge of being in VR, needing only to refer to my memory of putting on the headset and entering the program–and, as noted by Slater and Sanchez-Vives, such knowledge can “dampen down [participants’] responses.”
While it’s true that my anxiety to save the virtual people in the Trolley Problem was much milder than it would be in real life, it seems to me that the program’s inadequate presence-inducing capabilities–as opposed to my knowledge of being in a simulation–played the primary role in causing that disparity. There were significant weak points in the creation of both place illusion and plausibility, including a glitch that periodically made the virtual setting disappear to reveal my real surroundings. In the absence of such weak points, I doubt that understanding the artificiality of my environment would have, by itself, significantly influenced my responses. I was already fairly invested in this imperfect program, and if my presence had been magnified to a sufficient level, I imagine that the illusion of reality would have been strong enough to make me respond exactly (or nearly exactly) as if it were real. Still, as Slater and Sanchez-Vives point out, we can never guarantee how closely one’s virtual actions align with their analogous actions in reality (assuming we’re studying a situation where no “real-life” data can be collected), which means we cannot generalize VR-based data with absolute certainty. Nevertheless, when we simulate events that naturally occur in real life (rather than highly unlikely hypothetical scenarios like the Trolley Problem), we can “construct predictive theory that may help us understand how people might respond in reality” and later assess its validity by observing people’s behavior in real-life instances (Slater and Sanchez-Vives). Thus, despite their knowledge that the VR simulations are not “real,” officers can use them to predict their own behavior in the kinds of situations they will face (assuming the simulations have sufficient presence-inducing capabilities, which continued advancement should ensure). They can then make necessary adjustments before those predictions are put to the test.
The Emerging Use of VR in De-Escalation Training
In order for the police to effectively perform their duty–namely, to “protect and serve”–they must be well-equipped to prevent and avoid violence, whenever possible, through de-escalation. They must be prepared to use appropriate strategies and to overcome any emotions or attitudes (such as fear and bias) that may drive them to use unnecessary force. VR simulators like Apex Officer are designed to optimize such preparation. According to an article covering the use of Apex Officer by a police department in New London, Connecticut, the technology allows fellow officers to construct situations for trainees so that, “just like in real life, officers don’t know what they’re stepping into” (Hartz). This system ensures that the VR scenarios are sufficiently realistic (given their basis in real police experience) and novel to the trainees. Furthermore, VR training is “always followed by a debriefing session where officers receive feedback and consider what improvements they might make” (Hartz). The way trainees conduct themselves in a simulation may predict how they would act in real life at that point in their training, but these regular debriefing sessions provide the opportunity to shape their behavioral tendencies over time. Repeated practice with the simulator–during which officers “test out nonviolent strategies, try new ways of gaining voluntary compliance and improve decision-making” in a low-stakes environment where no one can be harmed (Hartz)–when accompanied by self-reflection and insight from outside observers, seems certain to lead to improvement. This depends, of course, on departments taking a thorough, serious approach to the training–but given the high cost of VR equipment, those who make the investment certainly have an incentive to do so.
According to the Apex Officer website–on which, despite the disturbing presence of the Blue Lives Matter flag in promo, the company appears genuinely dedicated to helping officers “make better decisions,” act with “appropriate discretion,” and treat “all people with respect and dignity”–thousands of law enforcement agencies, including Colorado State Patrol, Richmond PD, and New York State University PD, already use this particular simulator in their training programs. Presumably, in our capitalist economy, many competing products will emerge; competition will likely encourage technological advancement, which should give rise to increasingly immersive (and thus, increasingly effective) platforms. Therefore, police departments should continue investing in VR-based de-escalation training; the more widespread its use, the more widespread its benefits to new officers (and, by extension, the communities they serve).
Hartz, T. (2022, December 11). Conn. Police Using Virtual Reality to Train Officers in De-Escalation Tactics. Police 1. https://www.police1.com/police-products/virtual-reality-training-products/articles/conn-police-using-virtual-reality-to-train-officers-in-de-escalation-tactics-cqAX5CaI7FExerau/
Slater, M., & Sanchez-Vives, M. (2016). Enhancing Our Lives With Immersive Virtual Reality. Frontiers in Robotics and A.I., 3. https://doi.org/https://doi.org/10.3389/frobt.2016.00074