ARLINGTON, Va. — Just how well government agencies respond to future disasters may well depend on geeks and algorithms.
According to Gwendolyn Campbell, senior research psychologist at the Naval Systems Air Command (Navair),
two types of computer modelling could be the difference in avoiding or resolving a crisis.
Campbell and Donald Laughery, founder of Micro Analysis & Design, demonstrated at the Information Sharing and Intelligence conference late last week how the modelling types are being used to train soldiers and predict the resources needed to deal with an emergency. Campbell defined modelling as “”a way of representing regularities or patterns between input conditions, processes and outcomes.””
While modelling isn’t new — it’s been used for years to predict how a fire will spread, for example — cognitive modelling is still in its infancy, according to Campbell. When law enforcement officials storm a building or ship’s crew fighting a fire, you need to be able to accurately predict human processes.
“”What is the kind of performance you can expect out of people, the variance, the standard deviation? How’s that impacted by fatigue?”” Campbell said. “”It’s not enough to know that you have good hardware and good software to have an effective system of any kind. The people are the critical component, and you have to know how they’re likely to react.””
Campbell said there are a number of ways to build human performance models. It can focus on the sequence of tasks people are likely to fulfill, timing parameters, probability of error, what they do, how quickly and how well. It can also focus, she said, on how the mind functions. This will involve processing audio and visual information and examining how this data is interpreted to make decisions and ultimately take a course of action.
Understanding how people think and react is crucial when constructing a virtual environment populated with an enemy, Campbell said. Otherwise, the training is useless.
“”They have to make use of coverage and concealment. They don’t have a pre-programmed script. They have to work adaptively and flexibly. They have to react to noises down the hall,”” Campbell said. “”They have to communicate with each other.””
A smart opponent not only makes for better-trained soldiers and police officers, but it allows them to try new tactics, she added.
Laughery demonstrated process modelling, in this case a fire on a ship. Process modelling looks at human processes and behaviour during an emergency. The goal is to devise know how to structure a crew and arrange equipment on the boat to deal with the fire. It is also useful, he added, to determine how many people are needed to fight the fire and what strategies to use.
Laughery echoed Campbell’s opinion that these models are only as good as the algorithms can accurately predict behaviour.
“”Fatigue plays into these things. You don’t want to build a scenario or response which sort of assumes everyone’s bright and perky,”” Laughery said. “”These are all things you would want to see and use to learn how to improve. The purpose of doing all of this is to essentially try and find a better way.””