Why Faces Look Distorted in Our Periphery

In our lifetimes, we encounter hundreds of thousands of faces. Each person's unique exposure to faces largely determines their "face space"—a model of how we encode, perceive, and remember the faces that we see.

Cross-cultural differences, such as the other-race effect, demonstrate that we are generally more accurate at recognizing faces of ethnicities we are more familiar with. But even within cultures, people differ greatly in which faces will elicit a sense of trustworthiness, approachability, or attractiveness.

Another factor has an even greater effect on how we perceive a face—where the face is positioned relative to our central or peripheral vision. Something very strange happens when a sequence of faces is quickly presented in our peripheral vision. This was first demonstrated in 2011 in a powerful illusion discovered by Jason Tangen, Sean Murphy, and Matthew Thompson, called the "flashed face distortion effect" (see a demonstration below).

After a few seconds of staring at the central fixation cross, the faces peripherally presented faces start to appear deformed, cartoonish, and even grotesque. However, upon closer inspection, one can tell the faces are perfectly normal and have not been altered.

Several studies have attempted to explain the mechanisms underlying this phenomenon. Is it the fact that the faces are aligned at the eyes? Is it because they are presented quickly (4-5 faces per second)? Is it dependent on which faces follow each other or how centrally or peripherally they are shown?

The Role of Peripheral Vision

A study by Bowden et al. (2019) explored the role of peripheral vision. We know that everything in peripheral vision looks blurry compared to central vision. This is because peripheral stimuli are processed by retinal rod cells that have poor spatial acuity. So it could be that the intrinsic blur of peripheral vision makes the faces appear distorted.

However, Bowden et al. compared peripherally presented faces to centrally presented faces manipulated by blur. Although they found some evidence of face distortion in central vision using blurred faces, the effect was much larger when using un-blurred peripheral faces.

Another study by Balas and Pearson (2019) tested additional factors that may influence the illusion. They again found that peripheral presentation makes a big difference, as does the presentation speed.

Notably, the orientation of the faces presented did not seem to matter. In other words, upside-down faces produced a similar distortion effect as upright faces. This is surprising because, in most aspects of face perception, upside-down faces produce weak effects due to their general unfamiliarity (a prime example of this is the Thatcher Effect).

Do We All Perceive the Same Distortions When Looking at Peripherally Flashed Faces?

Earlier research by Afraz and Cavanagh (2009) found that peripheral distortion effects are quite idiosyncratic. Their study presented participants with faces and other stimuli in different positions of their peripheral vision. Participants were asked to make various judgments, such as the gender and age of faces or the aspect ratio of shapes.

They found that the very same stimulus can appear differently depending on where it is placed within the observer's peripheral vision. For example, a neutral face might appear male in the upper right corner, but in the lower right corner, the same face might appear female. These differences are stable over time for a particular observer but vary substantially between observers.

Therefore, the particularities of the visual distortions we experience in peripheral vision may be responsible for the particular ways flashed faces appear distorted. Further research is needed to identify the neural mechanisms behind these effects, whether they occur when we observe non-human faces, and whether they serve any adaptive purpose.