cognitive science
and more
Invisibly morphing faces

In the animation below, you can see a ring of four faces gradually morphing between Leighton Meester and Natalie Portman. If you keep your eyes fixated on the green cross in the center, you will notice that the morphing is quite difficult to see—but only when the faces are moving. If you match the movement with your eyes, or if the faces stop moving, you can clearly see the morphing happening.

This illusion was presented twice, in different guises, in the finale of last year's vision sciences illusion contest. Once by Jordan Suchow and George Alvarez, who ended up winning the contest. And once by Rob van Lier and Arnon Koning. So this illusion may not be entirely original, but it is certainly the most aesthetically pleasing rendition out there!

So what's going on here? Previously, I explained this phenomenon in terms of retinal motion. When the faces move, they slide across your retina (the light sensitive part of your eye), at least if you keep your eyes still. So the same face will be 'seen' successively by different parts of your retina. And because the organization of the retina is roughly preserved in visual areas of the brain, the same face will be successively processed by slightly different (sub)areas of your brain. According to the 'retinal motion' explanation, these different brain areas do not communicate effectively enough for changes to be detected. At least not if they are small and gradual.

But, as it turns out, this explanation is not correct!

The video below shows a counter-example. Here, you can see that the faces don't need to move for the illusion to occur. Change detection is also suppressed when the background is moving (like before, you need to fixate the dot in the centre). This effect is not as large as with the moving faces, and the background motion needs to pretty vigorous. But the illusion is definitely there!

So if retinal motion doesn't explain the illusion, what does? My guess is that this illusion has to do with something that you might call 'change adaptation'. Essentially, the idea is as follows: If nothing much is going on, the visual system is very sensitive to changes. Therefore, if neither the faces nor the background move, the subtle morphing readily pops out. But if a lot of things are changing at the same time, the visual system becomes less sensitive: It increases the amount of change that is required for us to consciously perceive a change. In a sense, we protect ourselves from being overwhelmed by too much change.

By analogy, you can think of what happens to your vision when you step out from the light into the dark. At first, you will hardly see anything. But you quickly adapt to the darkness, and vision is restored. And, just like lots of light will lead to adaptation of light detection, lots of change may lead to adaptation of change detection.

Of course, this is just an educated guess. The correct explanation may be something different altogether. But that's the cool thing about optical illusions: They are like puzzles, which you can solve by thinking of clever new variations.

References

Suchow, J. W., & Alvarez, G. A. (2011). Motion silences awareness of visual change. Current Biology, 21, 1–5. [Link: Free fulltext]

Suchow, J. W., & Alvarez, G. A. (2011). Background motion silences awareness of foreground change. In Special Interest Group on Computer Graphics and Interactive Techniques (pp.29-29). New York, NY: ACM. [Link: Subscription required]