cognitive science
and more
Boerke and the tell-tale pupils

Ok, stop procrastinating for a moment, and spend your time on something useful. Like finding Boerke in this image:

Boerke, shown at the bottom left, is hiding among his comic-strip buddies. (Source: Stripmuseum Brussels)

While searching for Boerke, you probably scanned the picture in a particular way: You scrutinized small parts of the picture one by one, looking at things that you wouldn't normally look at. Maybe Boerke is behind the ticket counter? (And where's the ticket counter?) No ... Maybe Boerke is climbing the stairs then? No, that's Bobette ... And so on, until you spotted Boerke. (Assuming you have. If not: keep looking!)

But this is not how you usually look at pictures. If you weren't searching for Boerke, you would probably let your eyes wander across the picture, focusing mostly on conspicuous things, like the dinosaur and the rocket.

So, broadly speaking, there are two modes of looking that differ in the amount of effort that you invest: An 'effortful' mode, in which you actively control where your eyes are going, and a 'lazy' mode, in which you let your eyes wander and be drawn to conspicuous things. And, as my colleagues and I have shown in a recent paper (public PDF), it is possible to determine how much effort you invest in your eye movements, simply by monitoring the size of your pupils.

Our experiments were similar to a search-for-Boerke game: Participants searched for a small letter that was hidden somewhere in an image. This was a difficult task that required effort. And when you invest effort in something, your pupils grow large—a basic physiological response. Therefore, we used pupil size as a measure of how much effort participants invested in the task.

Next, we tested whether people made different kinds of eye movements, depending on the amount of effort that they invested. More specifically, we tested whether participants looked at less conspicuous things when they invested more effort (i.e. when their pupils were larger). The logic behind this is that when you're in 'lazy-looking mode', your pupils are small, and your eyes are likely to get captured by conspicuous things: big things (e.g. the dinosaur in the Boerke picture), bright things (e.g. the fumes from the rocket), things that differ from their surroundings (e.g. the black box), etc. And to avoid your eyes from being captured by conspicuous things, you need to actively control where your eyes are going, which takes a bit of effort, and causes your pupils to grow large.

In other words, we tested whether large pupils, reflecting effort, predict fixations of the eye on inconspicuous things. And they do! Across three experiments, we consistently found a negative correlation between pupil size and the conspicuity of the things that people look at. Here's the relationship for one experiment:

The relationship between pupil size and the conspicuity of things that people look at. (Adapted from Mathôt et al., 2015)

So you can determine whether people are actively controlling their eye movements, and how this control fluctuates over time, by monitoring the size of their pupils. On average, of course, because we are dealing with a correlation that, while robust, is not very strong. Also, the fact that we're dealing with a correlation means that you cannot say for sure what causes what, and my description here is therefore an interpretation (but a plausible one, I think).

I think this finding is interesting, and possibly useful in real-life variations of a search-for-Boerke game. For example, take radiologists that search for abnormal patterns in medical images. Or airport-security personnel that search for suspect items in luggage scans. These are difficult tasks that require sustained effort—no dozing off on those jobs! And having some insight into the amount of effort that people are actually investing could be very useful.

References

Mathôt, S., Siebold, A., Donk, M., & Vitu, F. (2015). Large pupils predict goal-driven eye movements. Journal of Experimental Psychology: General. doi:10.1037/a0039168 (public PDF)