Participants were faster at detecting target faces with conspicuous sclera (white sclera or sclera colored lighter than the iris color) compared to faces with inconspicuous sclera (sclera colored similar to the iris color or darker)

Sclera color enhances gaze perception in humans. Jessica L. Yorzinski, Jacob Miller. PLOS, February 27, 2020. https://doi.org/10.1371/journal.pone.0228275

Abstract: Gaze perception is an essential behavior that allows individuals to determine where others are directing their attention but we know relatively little about the ways in which eye morphology influences it. We therefore tested whether eyes with conspicuous morphology have evolved to facilitate gaze perception. During a visual search task, we recorded the eye movements of human participants (Homo sapiens) as they searched for faces with directed gaze within arrays of faces with averted gaze or the reverse; the faces were large and upright, small and upright, or large and inverted. The faces had sclera that were conspicuous (white or colored lighter than the iris color) or inconspicuous (colored the same or darker than the iris color). We found that participants were fastest and most accurate in finding the faces with conspicuous sclera versus inconspicuous sclera. Our results demonstrate that eyes with conspicuous morphology facilitate gaze perception in humans.


Discussion

Our results support the hypothesis that eyes with conspicuous morphology have evolved to facilitate gaze perception. We found that eyes with conspicuous morphology are necessary for rapid gaze perception in humans: adult participants were faster at detecting target faces with conspicuous sclera (white sclera or sclera colored lighter than the iris color) compared to faces with inconspicuous sclera (sclera colored similar to the iris color or darker). These results demonstrate that conspicuous sclera, in contrast to the camouflaged sclera observed in most other primate species [6], are critical for rapid gaze perception in humans.

Human participants rapidly fixated target faces with conspicuous sclera. Regardless of whether participants were searching for the face with directed gaze among faces with averted gaze or the face with averted gaze among faces with directed gaze, they quickly found the target face with conspicuous sclera. In contrast, when the sclera was inconspicuous, participants spent more time searching for those faces compared to the faces with conspicuous sclera. These patterns were similar for the large faces, which simulated close-up interactions, as well as small faces, which simulated distant interactions. They were also similar for the inverted faces, which preserved low-level visual properties (such as luminance) but disrupted facial configuration [26]. Because visual attention can be strongly influenced by bottom-up processes (such as contrast; [27,28]), it is not necessarily surprising that participants were fast to fixate target faces with conspicuous sclera that exhibit high contrast. Importantly, participants were equally rapid at fixating faces with naturally-colored sclera and faces with sclera color that were lighter than the iris color; this demonstrates that digitally manipulating sclera color does not necessarily alter fixation latencies. We also found that participants were quicker to fixate target faces with light versus dark colored irises; future experiments can assess how iris color interacts with sclera color to influence gaze perception.

After participants fixated target faces with conspicuous sclera, they were quick to manually indicate that they had found the target faces (via a key press). However, when the sclera was inconspicuous, participants were much slower to manually indicate that they had found the target faces; their manual response times were especially slow for the small faces with sclera color that were darker than the iris color. These slower manual responses likely indicate that the participants were less certain of their decisions and therefore taking more time to indicate their responses. Their manual response times for faces with naturally-colored sclera and faces with sclera color that were lighter than the iris color were similar, indicating that digitally manipulating sclera color does not necessarily alter manual response latencies.

Participants were highly accurate in selecting the target faces. With the exception of the small faces with sclera that were darker than the iris color, participants were over 98% accurate in finding the correct target face. Therefore, even though participants spent more time searching for the target face when the face had inconspicuous sclera (see above), they were accurate in ultimately identifying the correct face. The only target faces that participants had difficulty in correctly identifying were the small faces with dark sclera color: they were 90% accurate in finding these target faces. Previous work found that humans are only 52% accurate in gaze perception when the sclera are darker than the iris (black sclera, white pupil, and white iris [7]). In contrast, we found that accuracy in gaze perception was over 90% regardless of sclera color. Rather than presenting stimuli in which the sclera and iris color were polar opposites [7], our stimuli only adjusted sclera color (rather than simultaneously adjusting sclera, pupil and iris color) and we did so relative to the natural iris color. Given that most primate species have sclera that closely match their iris color [6], our stimuli in which the sclera color matched the iris color closely simulated the eye morphology of these other primate species. Furthermore, our study used a gaze search task while previous work used a gaze discrimination task, which could also account for differences in accuracy levels.

Surprisingly, search efficiency was generally similar for faces with naturally-colored sclera and faces with sclera that matched the iris color. When the faces were large and upright as well as large and inverted, participants were equally efficient at searching for these faces. However, participants were less efficient at searching for faces with sclera that matched the iris color versus naturally-colored sclera when the faces were small, suggesting that sclera color in gaze perception may be especially important during distant interactions. The ability of humans to communicate with each other over large distances is likely critical to many types of interactions. This ability may be especially useful during a hunt when silent communication is critical for a successful outcome [5,29]. Previous work has shown that humans are able to reliably detect gaze direction from as far away as 10 meters [30,31] and our results suggest that sclera color contributes to this ability.

Our results support the gaze enhancement hypothesis by demonstrating that eyes with conspicuous morphology facilitate gaze perception. Humans are fastest and most accurate in gaze perception when faces have naturally-colored sclera (or sclera that are lighter than the iris color). When faces are modified so that eye morphology is less conspicuous (sclera that match the iris color or are darker than the iris color), gaze perception is slower. Given that 99% of nonhuman primates have eye morphology that is less conspicuous (with sclera color that closely matches their iris color; [6]), future experiments that examine nonhuman primates’ ability to perceive gaze would be informative. Our results demonstrate that eye morphology plays a critical role in human gaze perception and therefore impacts the evolution of social cognition.