To better understand human behavior, Penn State researchers in the School of Information Sciences and Technology (IST) have developed a computer model that simulates math anxiety when programmed to respond to a serial subtraction task.

Although in its simplest stages, the study will allow for greater insight on how and why humans react when they feel threatened by a situation, said Frank Ritter, associate professor of IST and psychology and head of the research team.

"With these models we can codify what we've known all along" -- that people do better on a task when they are challenged instead of threatened, Ritter said.

The data shows that when a subject feels threatened by a task, he or she will react more slowly. Correctness, however, will remain generally the same as to when threat is not an issue. When the subject feels challenged, he or she is likely to perform well.

"These results probably pertain to many other tasks," Ritter said.

By interpreting this data, researchers can develop corrective strategies to relieve stress, he said.

Next, the researchers will be studying how caffeine interacts with stress to improve performance. They will also look at its effects on blood pressure and heart rate, said Laura Klein, assistant professor of biobehavioral health and member of the research team.

"We are moving in the direction of looking at more complex tasks that require a more cognitive process. In the future we will be looking at gender differences and stress hormones," Klein said.

Klein said she expects results by the end of this year.

ACT-R, the program used for the study, is made up of cognitive mechanisms that are fixed for the task. These mechanisms can be considered as a set of instructions given to the computer that define how human perception and memory work, without any external factors acting upon them. Behavioral moderators that simulate human reactions, such as anxiety, are added to the model in order to evaluate the change in results. The moderators were compared to actual human data taken from the Journal of Personality and Social Psychology, according to the study.

Ritter compares the research to a computer game of chess. Unlike humans, computers will never give up and never feel threatened by their opponent. The researchers' job, instead of making the computer more intelligent, is to program the computer to be able to make sense of the task, Ritter said. In doing so, a game of chess would be more realistic.

This research can further be applied to video games and military simulations, he said.

The study says, "As research continues, ACT-R evolves towards a system that can perform the full range of human cognitive tasks: capturing in detail the way we perceive, think about and act on the world."

The researchers explained their data in a paper, "Using Cognitive Modeling to Study Behavior Moderators: Pre-task Appraisal and Anxiety," published this year and presented at the Human and Ergonomics Society 48th Annual Meeting in New Orleans last month.