Nicolas Davidenko¹ (), Mika Peer Shalem, Monique Crouse, James Pedroff; ¹University of California, Santa Cruz

Faces elicit a “left gaze bias”, or a tendency to fixate on the left side (often on the left eye). However, recent work has shown that tilting a face clockwise or counterclockwise disrupts this bias, revealing a different bias to fixate whichever eye is higher (termed the "upper eye bias"; Davidenko, Kopalle, & Bridgeman, Perception 2019). In the current study, we investigated whether the left eye and upper eye biases are face-specific, by measuring whether and how they manifest with and without a face context. We recorded participants’ eye movements while they reported whether a pair of symbols shown on a screen were asterisks (*) or crosses (+). In one block of trials, the two symbols were presented in the eye positions of parametric face drawings shown at a wide range of orientations (0, ±15, ±30, ±45, ±60, ±90, ±135, and 180 degrees). In the other block (counterbalanced order), the pair of symbols appeared in the same positions, but with no face present (i.e. on an empty background). The two symbols always matched, so participants could perform the task by looking at either one. An analysis of first fixations revealed (1) a strong bias to fixate the left symbol (observed in 16 of 19 participants) and (2) a strong bias to fixate the upper symbol (observed in all 19 participants). Critically, these biases were highly correlated and nearly identical in the face and no-face conditions (except for a slightly larger left bias in the face vs. no-face condition at some counterclockwise orientations). Overall, our results show that the “left eye” and “upper eye” biases manifest similarly whether there is a face present or not, suggesting that these biases are largely driven by general attentional mechanisms not related to face processing per se.

Acknowledgements: We thank Moorea Welch, Ghazaleh Mahzouni, Angela Huang, Joy Churchill, Lillian Nabkel, and Rishika Isanaka for their help collecting and analyzing data.