Decision-Making Under Fire

Lessons for training

Publication

Cook, R., Summers, D., Paulk, R., & Kirsch, E. (2022). Police officer responses to deadly encounters with the public: Understanding situational characteristics that impact decision-making. Policing: A Journal of Policy and Practice, 16(3), 355–369. https://doi.org/10.1093/police/paac026

What Was the Issue?

Cook and colleagues sought to understand how officers make decisions in deadly force encounters and what situational elements shape those decisions. At the broadest level, the study was motivated by ongoing public concern over officer-involved shootings and the recognition that small errors in judgment during high-stakes encounters can have long-lasting consequences for community trust. The authors framed two research questions: first, whether different types of scenarios would produce distinct patterns of outcomes, and second, which situational cues officers themselves identified as most important in guiding their responses.

How Did They Look at It?

To address these questions, the researchers recruited 39 sworn officers from Southeastern U.S. departments, the majority of whom were male, White, and averaged more than a decade of policing experience. Each officer completed twelve firearms training simulator (FTS) scenarios, six of which required deadly force and six of which did not. For this study, the analysis focused on the 233 encounters in which deadly force was necessary. The scenarios were presented on a VirTra V-100 LE simulator and designed to capture a range of real-world complexities. Officers used a Glock 17 modified to simulate recoil, and were free to move about in the projection space.

The six deadly force scenarios varied in structure and difficulty. They included a man concealing a firearm while appearing to set down a cinderblock, a car stop with a hidden gunman in the back seat, a workplace dispute escalating into a hostage situation, a nighttime dumpster encounter involving a screwdriver that could be mistaken for a weapon, a mentally ill subject threatening self-harm with a knife while a family member pleaded for restraint, and a stolen truck that involved multiple suspects, a hostage, and a language barrier. These scenarios deliberately incorporated environmental challenges, ambiguous objects, and multiple suspects in order to test how officers recognized and responded to unfolding threats.

Officer performance was categorized into five outcomes: passing the scenario (neutralizing the threat appropriately), missing the target, being shot first by the suspect, mistakenly shooting a victim, or shooting too soon. In addition to this behavioral coding, the officers were asked after each scenario to explain which situational factors had shaped their decisions. Their open-ended responses were analyzed through content analysis and coded into seven categories, including suspect cues, environmental factors, the presence of firearms or other possible weapons, suspect descriptors, impairment, and information from dispatch or bystanders. The coding process achieved acceptable levels of reliability, ensuring that the analysis reflected consistent interpretations across raters.

What Did They Find?

The overall results showed that deadly force encounters in the simulator produced a wide range of outcomes, with “pass” being the most common at about 45 percent of cases. However, nearly 30 percent of the time the suspect shot first, 13 percent of the time officers missed their target, nine percent of the time officers shot too soon, and about four percent of the time they mistakenly shot a victim. These aggregate numbers concealed important scenario-specific differences. For instance, in the cinderblock scenario, most officers were shot before reacting, and many admitted being distracted by the cinder block. In the traffic stop with the hidden gunman, nearly half the officers missed the target and many confessed they had not seen the second suspect emerge from the car. By contrast, in the workplace hostage situation, nearly 90 percent of officers passed the scenario, though one mistakenly shot the hostage.

Other scenarios proved especially difficult. In the nighttime dumpster encounter, more than a third of officers fired too soon, often misinterpreting a screwdriver or other object in the dark as a gun. In the mental health scenario, most officers were able to manage the situation appropriately until the subject charged with a knife, but about ten percent fired prematurely, citing the man’s manic state, the presence of a weapon, and the close distance. The final stolen truck scenario was the most challenging: fewer than one in four officers passed, 41 percent were shot first and more than 20 percent mistakenly shot the hostage. Officers themselves highlighted how multiple suspects, poor lighting, language barriers, and hostage dynamics compounded the difficulty of assessing the threat.

The qualitative data provided further insight into what shaped these outcomes. Across all six scenarios, officers most often mentioned suspect cues such as aggression, threatening actions, non-compliance, or erratic behavior as the decisive factor in their decision-making. Environmental factors such as time of day, bystanders, and the presence of multiple suspects were also frequently cited, as was the presence of a firearm. In contrast, suspect descriptors like race or gender, signs of impairment, or information relayed from dispatch or bystanders were much less frequently referenced. Many officers reflected candidly on being distracted by irrelevant elements, hesitating to fire when unsure, or rushing to act when an ambiguous object appeared threatening.

So What?

The authors argue that the findings highlight the difficulty of deadly force encounters even for experienced officers. The fact that nearly a third of scenarios ended with officers being shot first demonstrates the problem of hesitation under uncertainty. The relatively high rates of missed shots and premature firing highlight the perceptual challenges of quickly assessing ambiguous threats. Importantly, the study shows that decision-making is not simply about speed or accuracy with a firearm, but about perceiving, interpreting, and prioritizing situational cues in real time.

From a training perspective, the authors argue that the results suggest that exposure to a wider range of complex and ambiguous situations is essential. Officers need practice not only with marksmanship but with recognizing subtle suspect cues, distinguishing between harmless and lethal objects, and navigating environmental complications like darkness, crowds, or multiple suspects. Reflection exercises, in which officers articulate the factors that shaped their decisions, may help develop greater awareness of how distractions and misinterpretations occur. Finally, the scenarios underscore that errors are not uniform across situations; some contexts, such as hostage situations or language-barrier encounters, consistently produced greater difficulties. Designing training that replicates these high-risk conditions may improve officers’ ability to respond effectively in the field.

My $.02

I really want to highlight the authors’ argument that officers need exposure to a wider range of complex and ambiguous situations. I agree that this is important. Policing is an incredibly complex job and officers are called on to deal with an almost infinite variety of situations. Having a set of 5 or 10 scenarios that officers repeat periodically is simply not giving officers enough variation to help them develop the skills that they will need on the street. This is one of the reasons that I like the Constraints Led Approach (CLA). You can find a quick summary here and a really nice write up is here. When you use the CLA, you are constantly varying the constraints of the situation so that the trainees must learn to adapt to these changes. You aren’t forcing the training to rep out the “perfect” techniques hundreds of times. Instead, you are exposing them to hundreds of micro or mini-scenarios and they are learning to focus on what is important and take the actions that are needed to achieve their goals. In short you are making them problem solvers instead of technique reppers.

Finally, I want to comment on the limits of the technology. Simulators like the one in the picture above, limit the options that the trainee has. The trainee is constrained to stay in the general vicinity of where they started, but moving may give the trainee better options. In the picture above for example, if the officer moves to the left, he can limit the power that the suspect can put into throwing the block (because his hips are now square to the officer) and get a look at the passenger compartment of the truck. Of course, he wouldn’t be able to see in the bed of the truck any more, but this might be a reasonable tradeoff given that he has already seen in the bed. My point is not to say that one option is better than the other, but to illustrate that the option can’t be explored by the trainee in the simulator.