Reply to Gayet et al.: Minority salience as a social and cognitive phenomenon

In our original paper (1) we argued that social knowledge plays a meaningful role in minority salience (MS), partly basing our conclusion on findings from Experiment 5, where the effect of MS is larger when the minority of stimuli is composed of Black (vs. White) Americans. Gayet et al. (2) tested the hypothesis that the effect in Experiment 5 is driven by low-level characteristics of the display, not social knowledge.

In Experiment 1 Gayet et al. (2) examine whether a visual change to the background (white vs. black) moderates the effect of social knowledge. They report a three-way interaction, supporting the idea that visual aspects do play a role. Note, however, that the role of social knowledge in this experiment is best evidenced in the 2 × 2 interaction of ethnicity and minority. Crucially, it is significant [(F(1,185) = 8.17, P = 0.005], above and beyond background color. In other words, Gayet et al.’s (2) data indicate that social knowledge does play a role in the MS effect.

To further test this, we ran a preregistered replication of our Experiment 5 adding a white vs. black background condition (n = 200).* Like in Gayet et al.’s data, the ethnicity × minority interaction is significant [F(1,183) = 144.27, P < 0.001, η²_P = 0.44]. Moreover, this interaction is significant within both black and white background (P < 0.001; see Fig. 1). Given these patterns it is perhaps not surprising that the three-way interaction reported by Gayet et al. (2) is not significant. These results further support for the role of social knowledge in MS.

Fig. 1.

Results of our experiment. The zero line denotes accurate estimates, positive values denote overestimation, and negative values underestimation. In the Black American minority condition (red), the overall percentage of Black American faces was 25%. In the White American minority condition (yellow), the overall percentage of White American faces was 25%. The difference between the two minority conditions denotes social knowledge. This difference can be seen both on estimates of the minority group and on the estimates of the majority group, on both background conditions (black background or white background).

In Experiment 2, Gayet et al. (2) argue that the same pattern of results may be obtained with nonsocial stimuli (light vs. dark gray circles), suggesting that the MS effect is not social in nature. There are three points we wish to make here. First, the methodology of Experiment 2 is meaningfully different from Experiment 1 (and our Experiment 5), making direct comparisons difficult. Second, even if one ignores the first point, the authors report no quantitative comparisons to support their conclusion. Finally, qualitatively our original results, and those we report here, show a stronger MS effect for Black faces than for White faces (on a white background), quite the opposite direction of the pattern with circles (on white background). While it is not immediately clear how to interpret these differences, they seem to suggest circles do not behave like faces in this paradigm.

In conclusion, Gayet at al.’s data (2) and the new experiment we report here support the contention that social knowledge plays a meaningful role in MS and extend them by examining possible low-level vision effects in our results.

We thank Gayet et al. (2) for their interest in MS and for suggesting new questions that may improve our understanding of it. We are in complete agreement with them that such a basic and robust phenomenon will also be affected by purely cognitive factors, and we are looking forward to finding out more about the interaction between the social and the cognitive.