Publication
What Was the Issue?
Police officers sometimes must make split second decisions regarding whether someone is armed. Mistaking a common item, like a cell phone, for a gun can result in the death of an unarmed citizen. Failing to identify a weapon can result in the death of the officer. This study examines whether a brief training session can improve the speed and accuracy of weapon identification.
What Did He Do?
The author recruited 87 college students to participate in the study. These students were randomly assigned to one of two training groups. Both groups were pretested by having them complete 4 firearms training simulations using a MILO system. Two of these simulations contained armed suspects and two did not. Both groups then received three levels of progressively more complex training. The difference between the groups was that the experimental group was being trained to distinguish between common objects and weapons and the control group was being taught to distinguish letters (vowels from consonants).
During the first level training, the participants looked at a single picture and pressed a button to indicate whether it was a weapon or non-weapon in the experimental group. In the control group, they identified whether the picture was a vowel or consonant. For the second level, participants in the experimental group looked at pictures of an actor with objects of interest (weapons or non-weapons) somewhere on his body. The participants had to scan the actor, detect the object and then press a button to indicate that it was a weapon or non-weapon. Participants in the control condition saw letters and symbols arranged in a 4x4 grid. They had to scan the grid and then press a button to indicate that the grid either contained a vowel or did not. The third level of training increased complexity by using videos with movement in them. In the experimental group, the participants saw videos of actors turning around with an object in their hand or pulling an object out of their pocket. The participants had to press a button to indicate if the object was a weapon or not. In the control group, a camera panned over the 4x4 matrix, and the participants had to indicate whether the matrix contained a vowel or not. The total training time for both groups was about 20 minutes.
Following the training, both groups completed a post test that involved completing 10 scenarios using a MILO firearms training system. The actor in these scenarios was armed with a weapon in 4 scenarios, was holding a non-weapon in 4, and had no item in two scenarios. The order of the scenarios was randomized to control for order effects.
What Did He Find?
The researcher looked at three outcome measures. These were decision errors (e.g. indicating a gun when there was not), how fast they made the decision, and fixation speed (this is how quickly their eyes focused on the object of interest. This was measured using a vision tracker). The control group made more errors than the experimental (61 verses 22). This difference was large. The control group took 0.4 seconds on average to make their decision and the experimental group took an average of .36 seconds. While the experimental group was slightly faster than the control group in the raw data, the difference was not large enough to be confident that it was not a product of differences in who was assigned to what group. It took the control group an average of 0.46 seconds to fixate on the object and the experimental group took .39 seconds. This difference may seem small, but eye fixations happen very quickly and do not vary much, so this difference is statistically large and shows an effect for the training.
So What?
This study showed that a small amount of training can dramatically improve the ability of people to distinguish weapons from common items. Given the importance of making this distinction for officer and citizen safety, it seems that this type of training is worthwhile.
There is still a lot of work to be done in this area to try and further improve the training effect, assess how long it is retained, and ensure that it transfers to real world scenarios, but these initial findings are very promising. The researcher (Hunter Martaindale) is the Director of Research for ALERRT. If you are interested in helping him conduct more research in this area, please reach out.