At one point or another, we’ve all become frustrated by our computer, slamming our mouse down, aggressively pressing the keyboard. While people can tell if we’re angry based on the way we’re acting, the subtle movement and speed of our mouse can reveal our emotional state. Researchers at Brigham Young University (BYU) found people with negative emotions like anger, confusion, or sadness, become less precise in their mouse movements and move the cursor at different speeds.

“It’s counterintuitive; people might think, ‘When I’m frustrated, I start moving the mouse faster,’’ said Jeffrey Jenkins, study author, BYU professor and information systems expert, in the news release. “Well, no, you actually start moving slower.”

Jenkins and his researchers believe this behavior can be explained by the attentional control theory (ACT), which refers to an individual’s capacity to choose what they pay attention to and what they ignore. In other words, we go from focusing on what we want to the thing that’s upsetting us.

To determine how effectively mouse movement can predict emotional state, the researchers conducted a series of three experiments with dozens of computer users. In the first experiment, researchers randomly frustrated 65 participants from Amazon’s crowdsourcing marketplace Mechanical Turk as they performed a number-ordering task. By monitoring the participants’ mouse cursor movements, the researchers found frustrated users increased their cursor distance and decreased their speed.

Similar to the first experiment, the researchers randomly peeved 126 participants from a U.S. university, as they clicked on a mock e-commerce site. Upon monitoring their cursor movements, the researchers were able to identify when a user was frustrated 80 percent of the time.

In the third and final experiment, 80 participants from universities in Germany and Hong Kong reported their emotional levels as they interacted with a potentially frustrated online product configurator. The participants note their level of motion after completing each step in the configuration process. Here, the researchers found that cursor movement did not only give away negative emotions, but also the level of negative emotion.

Overall, the three experiments revealed frustrated people moved the mouse in more sporadic, jagged motions, while others moved the mouse more slowly. In comparison, a happy Internet user will move his or her mouse in a direct, precise manner to achieve a particular goal. A frustrated Internet user will be too distracted by what’s getting on their nerves.

These findings have future implications for the way we use our computers, specifically the Internet. “Traditionally it has been very difficult to pinpoint when a user becomes frustrated, leading them to not come back to a site,” Jenkins said. “Being able to sense a negative emotional response, we can adjust the website experience to eliminate stress or to offer help.”

A similar 2014 study designed a computer program that can accurately recognize users’ emotional states about 90 percent of the time, depending on the emotion. The researchers detected user emotions by analyzing the keyboard typing patterns of the user and the type of texts (words, sentences) typed by them. This combined analysis allowed them to detect various emotional states of the users, including joy, fear, anger, sadness, disgust, shame, and guilt.

The compilation of studies suggest interactions between humans and machines, like computers, could soon become ridden with emotional exchanges. For websites, being able to sense negative emotions can help web developers adjust the web experience to eliminate stress, or to offer help. It helps them to adapt or fix off points on their website that bring these negative emotions.

Jenkins’s technology has been patented and spun to a local startup company that holds the license. Currently, he’s in the process of fixing it, with details of his latest research appearing in top information systems academic journal MIS Quarterly.

So, will this technology be applied to mobile devices?

According to Jenkins, our swipes and taps cannot give us away — for now. But soon, Big Brother will not just be watching us, he’ll be able to read our emotions too.

Sources: Hibbeln M, Jenkins JL, Schneider C, et al. Inferring negative emotion from mouse cursor movements. Management Information Systems Quarterly. 2015.

Nahin NH AFM. Alam JM, Mahmud H, et al. Identifying emotion by keystroke dynamics and text pattern analysis. Behavior & Information Technology. 2014.