Nate Munk, Dr. Bonnie Anderson, and Dr. Anthony Vance, Information Systems

Habituation, the decreased response to repeated stimulation (Thompson and Spencer, 1966), to security warnings may be more serious than previously thought, due to generalization of habituation. Generalization occurs when the effects of habituation to a repeated stimulus carry over to other novel stimuli that are similar in appearance (Rankin et al. 2009). In this case, frequent system notifications generalize to less frequent security warning messages. For this project we are working with developers at Facebook and Google to develop and test warnings that resist generalization in order to make computer user’s more secure and to keep their information safe.

I had the chance to work on the design elements of the experiment from multiple perspectives. I went through and edited the draft of the experiment that was submitted to a conference. The paper was reviewed and returned with notes on how to improve the experiment.

I also worked through the potential designs of the experiment implementation. The purpose of revamping the experiment from the original idea was to create a realistic environment that gave the users tasks that were common for students. The original experiment showed various warnings and notifications throughout the user’s time working through it in a controlled environment. More recently we have been working on developing a way to have these warnings and notifications pop up while the user completes normal everyday tasks. Some of these tasks include checking email, looking at Facebook and working on an Excel spreadsheet. The tests were completed using eye-tracking tools.

We are working to develop a semi-realistic way to test the users in the future with eye-tracking and brain scans. fMRI gives us high spatial resolution, so we can see exactly which areas of the brain are affected. Eye-tracking tools have millisecond precision so we can partner the blood flood of the fMRI with the eye fixations measured with eye-tracking tools to pinpoint the exact stimulus and response pairing. In other words, by using both technologies concurrently, we can see how long a user looks at a certain space on the screen and how much brain activity is involved in the gaze.

Our prior research using fMRI to measure habituation in the brain in response to warnings showed a large drop in activity in the visual processing centers of the brain after only the second exposure to a warning, and found further decreases with additional exposures. Essentially this means that when a person sees a warning on their computer more than once they tend to start to ignore the warning. As we now look into generalization, we anticipate similar outcomes, showing the habituation to various computer notifications can cause users to tune out security warning messages. Once we determine how the effects of habituation to frequent notifications generalize to similar looking warnings, we will be able to measure the severity of generalization. We will work with developers (such as our contacts at Google and Facebook) to develop and test distinctive warnings that resist generalization.

This project is still ongoing and will continue into next semester. There is going to be a continuation of the technical design of the project to complete our objectives. I have worked closely with Dan Bjornn in figuring out how to implement PyGaze to create a more natural experience. We were hoping to be able to implement this previously; however, the technology only previously worked with webcams and had limited capabilities with better eye-trackers. PyGaze has recently released a tool that will allow it to work more accurately with Tobii eye-trackers. This recent technological improvement will allow us to create a more natural experiment.

We can increase our ecological validity if we are able to design more natural experiments. The goal in continuing the research is to better determine the effects of generalization of habituation, measure the severity of these affects, and continue to test new warning designs that can combat these issues.