An officer shoots an armed suspect, who later dies at the hospital. The officer is devastated but feels he did the right thing. The suspect’s family mourns and protests that the shooting was unnecessary. The community demands answers. Then comes the Chief’s press conference:
“Our investigation has determined that the shooting by Officer Smith was justified. This finding is supported by Officer Smith’s statement and the physical evidence. Ms. Suspect ignored the officer’s commands to drop the knife and continued to move toward Ms. Victim in a threatening manner. Ms. Suspect then stopped and lowered the knife. Officer Smith then ordered Ms. Suspect to get on the ground, but she ignored his commands. Instead, she raised the knife and started to lunge toward Ms. Victim, causing Officer Smith to fear that Ms. Victim’s life was in immediate danger. Officer Smith then fired his weapon three times to stop the threat. Ms. Suspect died from her injuries, but Officer Smith’s actions were necessary and saved the life of Ms. Victim.”
And then the video is released, a video of poor quality taken from a neighbor’s dated surveillance system. The police reviewed it as part of their investigation and did not see anything to contradict the statement of the officer and the physical evidence, including the relative locations of the knife and the victim. The public, and the attorney hired by the family to sue the department, see it as clear and compelling proof that Ms. Suspect was in fact moving away from Ms. Victim at the time of the shooting and did not lunge forward as stated by the officer. And, just to prove the point, the attorney releases “enhanced video” (zoomed in) which “clearly” shows the shooting is unjustified. Newspaper columnists watch the video and post their opinions that “it is clear from the video that the officer lied.”
Now the tragedy expands. The department and the officer are vilified in the press: The cops are trying to protect their own; they cannot be trusted to do their own investigations. The officer is labeled a liar. Years of efforts of developing good community relations are destroyed in weeks.
Kisela v. Hughes
The initial facts regarding this scenario are very loosely based on a 2018 decision of the United States Supreme Court, Kisela v. Hughes (138 S.Ct. 1148 (2018)), with a few additional facts to make certain points clearer. The press conference and fallout from the shooting are fiction but are also loosely based on several real incidents.
In Kisela, officers responded to a call of a woman, Amy Hughes, acting erratically and hacking at a tree with a knife. Upon arrival, officers saw Sharon Chadwick standing next to a car in a driveway. Hughes then came out of the house carrying a large kitchen knife and walked toward Chadwick. The officers had their guns out and were separated from the women by a chain-link fence. They commanded Hughes at least two times to drop the knife, but she kept it in her hand. Believing Hughes to be a threat to Chadwick, Officer Kisela fired at Hughes four times. The entire incident took about a minute.
Officers will not always have all the information needed to make an educated and controlled decision. Instead they must rely on their experience, training and what they do know about the situation.
Hughes survived her injuries and sued the department and the officers. The majority of the Supreme Court did not rule on whether Kisela violated Hughes’ Fourth Amendment rights. Instead they ruled Kisela should have been granted qualified immunity since, under all these circumstances, Kisela’s actions did not violate any clearly established law. Hughes was behaving erratically, possessed a large kitchen knife, moved within just a few feet of Chadwick, and refused commands to drop the knife. Kisela had to quickly make an assessment as to the potential threat to Chadwick based on the available information known to him at that time.
Two judges issued a strong dissent, ruling that a jury could find Kisela violated Hughes’ Fourth Amendment rights. They relied heavily on the fact that only one of three officers fired, while the other two subsequently stated they felt they still had time to attempt verbal techniques. The dissenting judges also pointed out Hughes made no aggressive or threatening movements; she was just close to Chadwick, holding a knife she would not drop.
There was no video mentioned in the case, but since this decision was issued, I have wondered how video of the incident would have affected the outcome. Would it help or hurt the officer’s case? What if the video showed Hughes was turning away from Chadwick at the moment she was shot? Would the ruling have been different? Would the video then be conclusive evidence the shooting was unjustified?
The Process of Thinking
A Google search of articles discussing the Kisela case revealed the following headline from the New York Post: “The Supreme Court just gave cops a license to shoot, then think.” While meant to be sensational, this idea reflects the science behind System 1 and System 2 decision making.
The concept of two different and distinct methods of decision making has been around for some time, but it became more widely understood and labeled with the publication of a book entitled Thinking, Fast and Slow by Daniel Kahneman. System 1 is fast, intuitive, emotional, automatic and heuristic-based. (Heuristics are mental shortcuts that we develop from previous experiences and problems.) System 2 is slower, deliberate and requires effort and attention.
Officers will be likely to rely on System 1 to respond to a rapidly unfolding, stressful situation. They will not always have all the information needed to make an educated and controlled decision. Instead they must rely on their experience, training and what they do know about the situation. Kahneman uses the phrase WYSIATI (What You See Is All There Is) to explain how we make decisions with System 1. We use the information available at the time. In potential life-or-death situations it may not be possible to engage System 2 by asking, “Hmmm, what do I not know?”
If the career or freedom of an officer rests at least in part on the detailed interpretation of video, the agency should hire a forensic video expert. Period.
In our scenario, the suspect is moving toward the victim with a knife, refusing commands to drop it. She stops and lowers the knife but then lifts it up again and moves toward the victim. At that point, the intuitive System 1 response would be a predictable and automatic reaction to what is perceived as an imminent threat. But what about the video that shows the suspect was moving away from the victim at the time she was shot?
Understanding Video Enough to Know You Do Not Understand Video
Understanding digital video well enough to use it for more than general observations is difficult. It is an incredibly complex area. But the little bit I do know makes this clear: If the career or freedom of an officer—or freedom of a suspect for that matter—rests at least in part on the detailed interpretation of video, the agency should hire a forensic video expert. Period.
For purposes of this article I will attempt to explain just enough to understand the issue pertaining to our scenario. I have had the pleasure of attending several classes on video given by Grant Fredericks, the founder and owner of Forensic Video Solutions. One of the first points Fredericks makes is that the interpretation of video evidence is almost always based on bias. Proof of this can be found at any Super Bowl party where fans of both teams are watching. Just wait for the instant replay of a controversial call and see who lines up on either side of the debate.
Beyond bias, there are many other limitations of video evidence in police use of force investigations, specifically with relying on video as an accurate depiction of what the officer experienced. A video (even a body camera video) does not show you what the officer was seeing at the time. It provides a much broader picture beyond what the officer can focus their attention on. This is particularly true with third-party video off a house or building, such as in our scenario, or video taken by a bystander. The perspective can be completely different, creating the danger of assuming the officer saw something that may have not been visible or obvious from his or her perspective.
All video uses compression as part of the encoding process, creating another issue when video is used as evidence. This compression process uses prediction to reduce the amount of data, which results in the fabrication of data. Nearly all compression algorithms are destructive, or “lossy,” meaning the lost data is generally unrecoverable. With DVRs, the compression rate may depend upon how long the data is to be stored. Also, file types such as AVI, MPEG, MP4 and WMV are all merely containers for digital data. The original proprietary file type unique to the source recorder is the best source of video. That is why video forensic experts say video that appears on YouTube is of no value forensically, because all metadata is gone.
While hindsight bias can be a factor in any use of force analysis, it becomes especially dangerous when video evidence is involved because we tend to assign video an unbiased perspective.
Another issue arises with interlaced video, meaning two streams of captured video are encoded as one image. Again, this is a process that allows for smaller data transfers and storage, but when there is movement in the video, interlacing artifacts can be created. As a result, blurred and distorted images may be created that do not accurately depict the true motion. Certain nuanced movements by a person cannot be recorded by video. For example, in our scenario, if the suspect had visibly clenched the knife as she brought it up, gritted her teeth and quickly moved her body and the knife toward the victim before turning, the video would not necessarily reflect these movements. The source recorder may not have the resolution, or possibly the angle, to pick up on those details, and the faster the motion, the more likely it is the motion will not be captured.
Finally, zooming in on a poor-quality image in a video only makes the image bigger, not better. This does not “enhance” the video.
The Impact of Hindsight Bias
I mentioned earlier that Fredericks underscores the effect of bias on video interpretation. Bias can be based on a person’s underlying beliefs and personal experiences. But there’s another bias that may be even more powerful, one the justices in Graham v. Connor tried to anticipate and prevent: hindsight bias. In our scenario, if Officer Smith had waited any longer, and the suspect was in fact intent on stabbing the victim, Officer Smith probably could not have stopped it. But in hindsight, that is not what happened and so there is a bias against the officer for not knowing at the time what the intent of the suspect was. What was learned after the shooting should be irrelevant to determining whether the officers’ actions were reasonable, but too often this knowledge becomes a factor.
While hindsight bias can be a factor in any use of force analysis, it becomes especially dangerous when video evidence is involved because we tend to assign video an unbiased perspective. The video has no reason to perceive things a certain way, after all. It’s not like an officer or a bystander or even an expert witness; it’s a machine. Fredericks and other forensic video experts can demonstrate that this is a flawed way to think about video, but the majority of the public continues to regard video as conclusive evidence.
The Graham factors language may not be enough to explain apparent contradictions between video evidence and the officer’s account or the outcome of the investigation.
The question that needs to be asked of the critics is, if the victim was stabbed while the officer did nothing, what would their attitude be then? If the victim was their sister, wife or daughter, what would they want the officer to do? The suspect was creating the situation, not the officer. Officers are thrust into such situations under high stress and must somehow correctly judge the intent of the person causing the risk. Officers must quickly judge what could happen, while the post-incident critics have the benefit of knowing what did happen.
In the Kisela case, the dissenting judges relied on the fact that only one of the three officers shot as a reason to doubt the reasonableness of the force used. This is interesting because it begs the question, if all three officers had fired multiple rounds at Hughes and killed her, would the judges then have used that to find their actions excessive? If the officers had waited for indisputable proof that the suspect was in fact a threat, instead of just being reasonably perceived as a threat (which, by the way, is the law under the Fourth Amendment), and the victim was killed and then the suspect was shot, the tragedy would have been doubled. But hindsight bias can preclude that analysis.
Putting It All Together
When preparing press releases and briefings, police leaders need to be aware the limitations of video evidence in police use of force investigations. The language used in the press conference in our scenario is typical. While it is not wrong, it also may not tell the complete story. Using the standard Graham language—“at that point, the officer feared for his or her (or third party) life and the officer fired to end the threat” or some variation—can be misleading. It implies the officer had time to make a conscious decision and was in control of the situation. It also leads to the implication the officer should have known the suspect was no longer a threat.
This will not always be the case. Instead, the officer is relying on the subconscious and automatic process of System 1, which is reacting to all the information known to the officer at that time. The subtle and nuanced movements and facial expressions of the suspect, which may not be visible on a video, become subsumed within the rapid processing of System 1. Rapid and brief movements of the suspect may not be captured by video because of a different visual perspective or compression techniques that distort or blur images. As a result, video may not be conclusive evidence and instead could be misleading.
Police leaders also need to be aware of the concept of hindsight bias and understand that perception of video evidence can be affected by knowledge of the outcome of the incident. Many police executives and officers have learned to use Graham factors language when discussing use of force incidents. That’s good because it aligns what happened with established case law regarding use of force, but it may not be enough to explain apparent contradictions between video evidence and the officer’s account or the outcome of the investigation. It is impossible to suggest a one-size-fits-all approach, but some general considerations can assist police leaders when making a public statement following an officer-involved shooting. Some of these may or may not apply to your situation, but generally you should:
- Understand and accept that you cannot do this all yourself. Develop subject matter experts and trainers within your ranks and consult with them. Encourage debate and discussion about possible issues. If necessary, reach out to other agencies that have experts you may not. Thinking that you know everything you need to know can make things worse.
- Emphasize your explanation of the “high stress and rapidly unfolding” aspect of the situation. Many use-of-force situations require officers to act automatically and responsively (System 1). Care must be taken to not create belief that the decision was conscious and deliberate if it was not.
- Address the dangers of hindsight bias head on. Keep the focus on the factors known by the officer(s) at the time of the shooting and remind the audience that anything learned afterward is irrelevant to the determination of whether the shooting was justified. If appropriate, discuss what could have happened but for the actions of the officer.
- Educate your audience about video’s limitations in perspective and image quality, but not just in general terms. If there are issues with the video from your incident, then both the technical aspects (e.g., method of compression) and the practical aspect (e.g., how that compression distorted an image) must be explained. If you use a forensic video expert during your investigation, ask them to help you develop these talking points.
- If applicable, emphasize that it is the suspect’s behavior and refusal to listen to commands that drove the incident.
- Do not make premature statements of fact before all the evidence has been reviewed.
- Do not wait for a tragedy to occur. Try to educate the public about these issues as much as possible before an incident occurs.
All the issues discussed in this article are assuming that the investigation finds the officer was justified in his or her response. If this is not the case, then police leaders face a much different conversation and of course, must take appropriate disciplinary action.