Graduate Student and Professor
win Statistics Award

 

Dr. John Braun and Jennifer Asimit

Dr. John Braun and Jennifer Asimit help unravel the link between stimuli and perception

By Mitchell Zimmer

Imagine driving down the 401 when the brake lights of the car ahead of you suddenly flash. In response, you apply the brakes as quickly as you can. However, in the slight delay starting from the time the image of the brake light reaches the back of your eyes to the time you push the brake pedal you are hurtling towards the car ahead. (Since this scenario is in your imagination, let's say you stop in time.) This matter of reaction time has been the subject of intense study by researchers in the field of visual psychophysics in an effort to learn about the brain mechanisms underlying vision. The above example provides a window enabling researchers to peer inside the complex operations the brain uses in performing a visual-motor task.

In the laboratory setting, experiments based on the similar “see a flash and push a button method” can become more complicated. There are occasions when two light flashes occur very close together and the participant only records one button push. A paper published in The Canadian Journal of Statistics by PhD student Jennifer Asimit and Associate Professor John Braun of the Department of Statistical and Actuarial Sciences investigated this problem. As Braun says, “In looking at a sequence of flashes and a sequence of button presses you now no longer have a one to one relationship. There is supposed to be a match between the flash and the button press but invariably you’ll have fewer button presses than flashes.”

“The data type that comes out is much more difficult to deal with because you have these gaps, basically missing data and you no longer know which flash is matched to which response. At the same time though, you get the opportunity to observe interactions among the flashes and responses that you wouldn’t see in a discrete single flash-single response experiment. With the work that Jennifer and I did in the paper we actually explored the interaction effect among pairs of flashes.” Asimit and Braun asked if there was there a pattern of response when two flashes were close together. “Could you predict whether you would have one or two or more responses? So we developed methodology that, in theory, could actually detect such patterns for the particular experimental data that we had. The result was a little disappointing, it was actually inconclusive. We couldn’t tell what the mechanism was other than that it wasn’t exactly what we postulated in our simple model.”

Despite the result, the method provided an exiting new inroad to this kind of research. “The problem in this reaction time experiment is determining, in the diminishing lag time between two flashes which are close together in a series, what flash elicits the pushed button the first flash or the second flash” says Braun. “From the methodology that we were working from you could actually compute a graphic which would show you very explicitly which flash was actually seen. To me, that was kind of amazing -that from a run of flashes and responses you could actually tell what somebody could perceive.”

As a result, both Asimit and Braun have been awarded the 2005 Canadian Journal of Statistics Award, the top honour presented yearly by the Statistical Society of Canada for excellence in statistical research.

The award recognized the outstanding quality of the paper's presentation and methodological innovation. The paper is titled "Third order point process intensity estimation for reaction time experiment."

 

Also from this web page: