Belief in biological origin of race (racial essentialism) increases sensitivities to cultural category changes measured by ERP mismatch negativity (MMN)

The dynamic multicultural view explains culture-specific effects on cognition that cultural knowledge is organized in multiple knowledge systems which are specific to each culture and differentially activated when exposed to related objects or scenes. This view predicts automatic categorizations of environmental information according to the culture-specific knowledge systems. This study investigated whether cultural information could be spontaneously categorized, and the modulation of this process by the belief in the biological origin of race (i.e., racial essentialism) with an event-related brain potential, the visual Mismatch Negativity (vMMN). Deviant pictures of Eastern (Western) culture were randomly presented in a stream of standard Western (Eastern) pictures while participants were playing a video game. Participants who endorse racial essentialism (high group) showed vMMNs to the deviants with high relevance to the Eastern or Western culture and the deviant with low Eastern relevance; while participants with low racial essentialism showed vMMN to the deviant with high Eastern relevance only. These results revealed spontaneous cultural categorization with vMMN and the top-down modulation of spontaneous categorization by personal belief. In addition, this is the first demonstration of MMNs to cultural deviance and the potentials in applying MMNs to study psychological essentialism and social categorization.


Linear Mixed-effect Model Analysis
The linear mixed effect model was used to analyze the effects of racial essentialism, as a continuous variable, while both cultural relevance (high cultural relevance coded as +1 and low cultural relevance coded as -1) and culture type (Eastern culture coded as +1; Western culture coded as -1) as categorical variable, and their interaction effects. Specially, were not statistically significant. The two-way interaction effects between racial essentialism and cultural relevance (beta = 0.23, t(152) = 0.59, p = .56), and between racial essentialism and culture type (beta = 0.64, t(152) = 1.67, p = .097) were not statistically significant. However, the two-way interaction effect of cultural relevance and culture type (beta = 4.59, t(152) = 2.40, p = .018), and the three-way interaction effect of racial essentialism, cultural relevance, and culture type (beta = -1.18, t(152) = -2.17, p = .031) were statistically significant. The vMMN response pattern and the results from the linear mixed effect model analysis were similar to that of the mixed-effect ANOVA shown in

Positive Mismatch Response / MMR
Most vMMN studies reported negative mismatch responses 1 , but some studies reported positive mismatch responses 2,3 (MMR). Since no prior study examined the vMMN to cultural changes, the positive responses at the Cz electrode in the vMMN time window were analyzed with procedures similar to that in analyzing the negative mismatch responses, except for measuring the positive instead of the negative response. One-sample t-tests showed significant positive mismatch responses to Western deviants with high cultural relevance in the low racial essentialism individuals (t(19) = 2.11, p = .024, d = 0.50) and Western deviants with low cultural relevance in the high racial essentialism  Assuming that the automatic detection system is less sensitive to the deviants with low cultural relevance and the Western deviants, as the participants were more used to the Eastern culture, the positive mismatch response elicited by the Western deviant with low cultural relevance could indicate an ambiguous or more difficult to-be-detected deviant.
This proposal is consistent with the positive mismatch response observed in the Western deviant with high cultural relevance in low racial essentialism individuals. Previous studies repeatedly demonstrated differences in the spatiotemporal dynamics of the fronto-sensory cortical network in detecting deviants with various degree of deviance [4][5][6][7] . The detection of ambiguous deviants typically involved additional early frontal mismatch responses than detecting salient deviants 5,7,8 . The association of the mismatch response polarities and sensitivities of the brain in detecting cultural changes needs to be clarified in future studies.

Frontal P3 as Alternative Explanation
At the Cz electrode (

6 Electrodes vMMN Results
For the vMMNs averaged across the 6 frontocentral electrodes, the mixed-effect ANOVA showed a significant main effect of culture type (F(1, 38)   Follow-up repeated measure ANOVAs with the factors cultural relevance and culture type were carried out on the vMMNs for the high and low racial essentialism groups, separately. For the low racial essentialism group, the main effects of culture type and culture relevance were not significant (Fs(1,38) = 1.12 and 0.03, ps= .31 and .88, η 2 ps = .06 and .001), while the two-way interaction effect between culture type and cultural relevance was significant (F(1,38) = 4.72, p = .043, η 2 p = .20). There was no significant difference between the vMMNs to the Eastern and Western deviants with low culture relevance (t(19) = 0.88, p = .39 , d = 0.28). However, the vMMNs to the Eastern deviants with high culture relevance was larger than that of the Western deviants with high culture relevance (t(19) = -3.15, p = .003, d = -1.00).

14 Electrodes vMMN Results
For the high racial essentialism group, the main effects of culture type, culture relevance, and the two-way interaction between culture type and cultural relevance were not significant (Fs(1,38) = 0.31 to 3.46, ps = .08 to .59, η 2 ps = .016 to .15). No significant differences were found between the vMMNs to the Eastern than the Western deviants with both low and high culture relevance (ts(19) = -1.46 and -0.54, ps = .15 and .59, ds = -0.46 and -0.17).

Non-parametric Randomization Test
Alternatives procedures for localizing the vMMN are available. For example, the vMMN can be localized from peak response of the grand average across all conditions. However, this measurement procedure is more appropriate for identifying the peaks of amplitude modulated components in the evoked ERP responses (e.g., increase in N1 amplitude of attended versus unattended conditions). This measurement procedure assumes that the ERP responses to the conditions are different in the amplitudes and has limited tolerance on the latency variations between conditions. As the current study was the first study on vMMNs to cultural deviants, and the latency variation between conditions due to individual differences of racial essentialism were unknown, this procedure may not be appropriate.
To examine whether the results of the current study was produced by the peak vMMN amplitude measurement procedure which could be more sensitive to noise, a nonparametric randomization analysis similar to the simulation procedure in Luck and Gaspelin (2017) was carried out. In this analysis, the racial essentialism group labels (high versus low) were shuffled among the participants, while the cultural relevance (high versus low) and culture type (Eastern versus Western) labels were shuffled among the difference waveforms of different conditions within each participant. The peak mismatch responses were measured from the grand averaged difference waveforms (across participants' waveforms of each condition) using the peak measurement procedure described in the