Who are you looking at? The influence of face gender on visual attention and memory for own- and other-race faces

Abstract

Previous research suggests that the own-race bias (ORB) in memory for faces is a result of other-race faces receiving less visual attention at encoding. As women typically display an own-gender bias in memory for faces and men do not, we investigated whether face gender and sex of viewer influenced visual attention and memory for own- and other-race faces, and if preferential viewing of own-race faces contributed to the ORB in memory. Participants viewed pairs of female or male own- and other-race faces while their viewing time was recorded. Afterwards, they completed a surprise memory test. We found that (1) other-race males received the initial focus of attention, (2) own-race faces were viewed longer than other-race faces over time, although the difference was larger for female faces, and (3) even though longer viewing time increased the probability of remembering a face, it did not explain the magnified ORB in memory for female faces. Importantly, these findings highlight that face gender moderates attentional responses to and memory for own- and other-race faces.

Keywords:

• Face recognition
• Visual attention
• Own-race bias
• Own-gender bias
• Sex differences

Acknowledgements

This research was supported by a grant from the Swedish Research Council awarded to Agneta Herlitz. We are grateful to Elin Frögéli for assisting with the data collection.

Notes

¹In a random intercept logistic regression model (e.g., Wright & London, 2009), participants’ “yes” and “no” recognition responses (yes = 1, no = 0) were regressed on the face being a target or a lure (faceold, 1 = old, –1 = new), race, gender, sex of viewer, and all interactions between these predictors. There were 2080 data points (n = 52, 40 data points per participant). The first model confirmed the ANOVA findings. For the effects of interest, controlling for facial attractiveness did not reduce the magnitudes of the unstandardised regression coefficients (B _model1/B _model2: faceold, 0.85/0.87; race×faceold, 0.21/0.22; gender×faceold, 0.23/0.23; race×gender×faceold, 0.12/0.11; gender×sex of viewer×faceold, 0.23/0.23) or change the significance levels (p _model1/p _model2: faceold,<.001/<.001; race×faceold,<.001/<.001; gender×faceold,<.001/< .001; race×gender×faceold, .01/.02; gender×sex of viewer×faceold, .02/.02). A face rated as more attractive had higher odds of being reported as seen before than a face rated as less attractive (B = 0.22, SE = 0.11, p = .046).