Lateralization of facial emotion processing and facial paresis in Vestibular Schwannoma patients

Abstract Objective This study investigates whether there exist differences in lateralization of facial emotion processing in patients suffering from Vestibular Schwannoma (VS) based on the presence of a facial paresis and their degree of facial functioning as measured by the House Brackmann Grading scale (HBG). Methods Forty‐four VS patients, half of them with a facial paresis and half of them without a facial paresis, rated how emotive they considered images of faces showing emotion in the left versus right visual field. Stimuli consisted of faces with a neutral half and an emotional (happy or angry) half. The study had a mixed design with emotional expression (happy vs. angry) and emotional half (left vs. right visual field) of the faces as repeated measures, and facial paresis (present vs. absent) and HBG as between subjects’ factors. The visual field bias was the main dependent variable. Results In line with typical findings in the normal population, a left visual field bias showed in the current sample: patients judged emotional expressions shown in the left visual field as more emotive than those shown in the right visual field. No differences in visual field bias showed based on the presence of a facial paresis nor based on patients’ HBG. Conclusion VS patients show a left visual field bias when processing facial emotion. No differences in lateralization showed based on the presence of a facial paresis or on patients’ HBG. Based on this study, facial paresis thus does not affect the lateralization of facial emotion processing in patients with VS.

evidence appears to suggests that the right hemisphere generally plays a more important role in emotion processing than the left hemisphere (Murray et al., 2015), evidence is definitely not conclusive and it is suggested that the two main hypotheses regarding the hemispheric lateralization of emotion processing are not mutually exclusive . Therefore, though this study is mainly focused on examining possible differences in lateralization of emotion processing between VS patients with and without facial paresis, we will examine the overall lateralization-in line with the right hemisphere hypothesis-as well as possible differences in lateralization based on valence-in line with the valence hypothesis.
The current's study addresses the lateralization of hemispheric processing by a method that has been extensively used in previous research: The chimeric faces test, a behavioral test of facial emotion processing which presents a face with an emotional expression in one half of the face and a neutral expression in the other half of the face. The image of the face is presented centrally, with the emotional facial expression thus being presented either in the left or the right visual field. This test examines whether there exists a bias in the observer considering the perception of emotional expressions presented in the left compared to the right visual field (e.g., Bourne, 2010;Bourne & Gray, 2011;Levy, Heller, Banich, & Burton, 1983). Hemispheric lateralization of emotion processing concerns the bias people tend to show in perceiving emotional expressions shown in the left or the right visual field as more emotional, or to recognize them more accurately depending on the visual field in which they are portrayed (Bourne, 2010;Murray et al., 2015).

Considering that the information that is shown in the left visual field
initially is received and processed by the right brain hemisphere, a left visual field bias is interpreted as support for the notion that the right hemisphere is more strongly involved in emotion processing than the left hemisphere (Bourne, 2006).
The role of the facial muscles of the observer in relation to hemispheric lateralization has been partly examined in healthy individuals as well as patients with mild unilateral facial paralysis (Blom, Aarts, & Semin, 2019;Korb et al., 2016) First, a recent study (Blom et al., 2019) using the chimeric faces test reported typical left visual field bias on perceived emotionality, but this visual field bias did not directly emerge in facial muscle activation. Furthermore, a study testing patients with acute, subacute or chronic unilateral facial paresis found that patients with a left versus right facial paresis processed emotional expression of happiness and anger equally. Interestingly, patients with a left facial paresis processed happy expressions more accurately when presented in the right versus left visual field, indicating a somewhat complicated relationship between facial paresis and emotional processing of others' expressions (Korb et al., 2016).
In short, although suggestive, the research conducted so far does not give a clear picture about the role of facial muscles in perceiving emotionality in facial expressions of others.
The current study aims to enhance the understanding of the possible role of facial mimicry in perceived emotionality by examining the impact of being limited in one's facial functioning on emotion processing of hemispheric lateralization. First of all, while the left visual field bias-in line with the right hemisphere hypothesis-has often been observed in healthy individuals, we aim to replicate this typical bias effect in a sample of patients with VS. Additionally, we test for possible differences in bias based on the valence of the emotional expression. If the patients show a left versus right visual field bias for positive versus negative facial expressions, this would relate to the valence hypothesis. Most importantly, however, we examined the role of facial functioning in hemispheric lateralization of emotion processing by comparing VS patients with and without facial paresis, as well as by examining the association between hemispheric lateralization of emotion processing and the degree of facial dysfunction as measured by the House Brackmann Grading scale (HBG; House, 1985). If facial functioning plays an important role in this, patients' facial functioning should be related to the visual field bias.

| Study overview
We investigated the role of facial functioning in how emotional patients with VS perceive faces showing emotional expressions in the left or right visual field, with the other visual field being neutral in expression. Treatment of VS can include surgical removal of the tumor that causes a degree of (chronic) unilateral paresis in the patient.
To take this important facial functioning difference into account, the study had a mixed design with emotional expression (angry vs. happy) and emotional half (left vs. right visual field) of the stimulus as repeated measures, with facial functioning (patients with or without facial paresis) as the main independent variable. The study was conducted and written informed consent of each participant was obtained in compliance with the principles contained in the Declaration of Helsinki. Permission for the study was granted by the Medical Ethics Committee of the Leiden University Medical Center.

| Participants
Incidence rate of VS is low, with an estimated incidence rate of 15 persons per million in the Netherlands-where the current study took place-, with the highest latest incidence rate in one specific region of the Netherlands being 33.2 (Kleijwegt, Ho, Visser, Godefroy, & van der Mey, 2016). Clearly, the number of VS patients experiencing a chronic condition of facial paresis due to surgical removal of the VS is even much lower. Considering this low incidence rate, we aimed at including a reasonable number of VS patients with or without facial paresis (N = 44) to examine interaction effects within our mixed design with two within subject repeated measures. Running a sensitivity analysis in G*Power 3.1 (α = 0.05, power = 80%, N = 44) for an ANOVA: Repeated measures within-between interaction (including the moderator test of the patient group as well) indicated that we were able to detect a small to moderate effect size, f = 0.18. Patients with and without facial paresis were matched as closely as possible (see Table 1 for details of the two subsamples) on the factors biological sex, age, side of the VS and the time that had elapsed since their diagnosis. In total, 28 females and 16 males participated (M age = 54.39 years, SD = 7.41 years). Twenty-two patients experienced a degree of facial paresis after removal of their VS, while twenty-two patients had a VS but had not developed a facial paresis. Seventeen patients had a VS in the right cerebello pontine angle, while twenty-seven patients had it in the left cerebello pontine angle.
The average time that had passed since being diagnosed with VS was 6.55 years (SD = 4.74). Facial dysfunction was graded by means of the House Brackman Grading scale (HBG); currently, the most commonly used and accepted scale to document patients' degree of facial dysfunction (Zandian et al., 2014). This scale contains six levels of facial nerve function, with a higher grade representing stronger facial dysfunction. The HBG was scored both by the experimenter and by the patients themselves. Inter-rater reliability was high: Pearson's r = .87, therefore, the average HBG was used for analyses.

| Participant recruitment and response rate
Patients applied for participation either via responding to a letter of invitation received from their treating physician, or via responding to a call for participants on an online forum for people with VS. 1 Out of the 62 patients who applied either via the online forum or who were invited by their treating physician, 42 (70.79%) decided to participate in the current experiment.

| Stimuli
Chimeric faces that were created for and used in an earlier study (Blom et al., 2019) were used in this study as well. The chimeric faces were generated using images of four female and four male faces from the Dutch Radboud Faces Database (Langner et al., 2010). Each chimeric face was composed of an emotional (angry or happy) half face and a neutral half face (see Figure 1 for an example) of the same model, by blending the faces at the midline. We used both the original pictures and the mirrored pictures. The effects that we would find would then thus be due to a true visual field bias, not to a possible difference in

| Procedure
Patients were informed that they had to rate on a 9-point scaleusing the numeric keys 1 to 9-how emotional they found each face presented to them on the screen. They used their preferred hand to give their response and were asked to not think too long about their rating and to trust their first impression. After four practice trials, in which patients could get accustomed to the task and to the type of images, the experiment started. The task was presented in two blocks, each block consisting of the same 64 trials, presented randomly without replacement. Each trial started with a blank screen (1,000 ms), after which a fixation point appeared (random time between 600 and 1,000 ms). Then, the chimeric face appeared with the rating scale below the face, which remained on screen until the face was rated. Patients went through the experiment self-paced and could take a break in between blocks if they felt the need to.
Average ratings of emotionality were calculated per stimulus type and served as dependent variable.

| Statistical analyses
We will first test the hypothesis that VS patients with and without

| Visual field bias and facial paresis in VS patients
In order to examine how patients respond to the chimeric faces, we For the sake of clarity of reading, below we first report the main effects, followed by all higher order interaction effects.

| Three-way interaction effects
The interaction between emotional half and valence did not show to be qualified by a further interaction with facial paresis, F(1, 40) = 0.15, p = .704, η p 2 = 0.00. A Bayesian independent samples ttest revealed that the data were 3.03 times more likely to reflect this null effect (BF 01 = 3.03), than for it to reflect a difference in visual field bias based on valence between the two patient groups.

| Four-way interaction effect
Lastly, the interaction between emotional half, valence, facial paresis, and side of VS was not significant, F(1, 40) = 0.06, p = .806, η p 2 = 0.00. The Bayesian analysis of variance indicated that the model including this interaction did not explain the data well compared to matched models not including this effect (BF incl = 0.39).
To conclude, while the classic left visual field bias showed for the current patient sample, VS patients with and without facial paresis did not show a difference in this visual field bias ( Figure 2).
Furthermore, this left visual field bias showed to be slightly larger for happy than for angry chimeric faces.

| Visual field bias and degree of facial dysfunction in VS patients
Second, it was examined whether the degree of facial dysfunction as measured by the average HBG score showed to be related to the above reported visual field bias. A score representing the The association between HBG score and visual field bias score for happy chimeric faces was not significant, F(1, 42) = 0.00, p = .964, R 2 = .00. HBG thus did not predict the visual field bias for positive expressions b* = 0.04, t(42) = 0.05, p = .964, B = 0.00, 95% CI B [−0.14, 0.15]. A Bayesian correlation revealed that the data were indeed 5.32 times more likely to reflect a null effect (BF 01 = 5.32), than for it to reflect an association between degree of facial dysfunction and the visual field bias for positive expressions.
The association between HBG score and visual field bias score for angry chimeric faces was also not significant,

| VS patient sample versus a healthy control sample
While we report a strong replication of the left visual field bias in A Bayesian ANOVA confirmed that the data were 7.82 times more likely to reflect a null effect (BF 01 = 7.82), than for it to reflect a difference in visual field bias comparing healthy controls, VS patients with facial paresis, and VS patients without facial paresis. Accordingly, we consider it most likely that the null effects were due to the absence of a relationship between facial functioning and lateralization of facial emotion processing, and not because of an affected visual field bias in the VS patient sample as a whole.

| D ISCUSS I ON AND CON CLUS I ON
The current study was aimed at examining hemispheric lateralization of facial emotion processing by means of the chimeric faces test in Vestibular Schwannoma patients with and without facial paresis.
First of all, we replicated the left visual field bias in this patient sample, meaning that when an emotional expression was depicted in the left visual field, rather than in the right visual field, the face was perceived as being more emotional. This left visual field bias showed to be somewhat stronger for positive (happy) than for negative ( facial paresis were able to detect in which visual field a happy facial expression first appeared. Though speculative, then, differences in patient groups and task measurements might have produced different findings between the studies as a result of tapping into different aspects of emotion processing (e.g., detection of emotion in faces vs. perceiving emotionality in faces).
Considering that the current study does provide a strong rep- Extrapolating those findings to the current study would suggest that hemispheric lateralization of facial emotion processing might be a process that relies more on visual and subcortical information processing, rather than on sensorimotor information processing involved in simulating the facial expressions of others.
Though our findings could be interpreted as evidence against the role of facial mimicry in emotion processing, we would like to stress here that the findings reported in this study do not necessarily go against the important function of facial mimicry. Other information-such as the visual (de la Rosa et al., 2018)-can sometimes provide sufficient input in order to complete emotion processing tasks, hence reducing the "need" for facial mimicry for certain tasks (e.g., Arnold & Winkielman, 2019). For example, while facial mimicry did show to relate to the valence of the chimeric faces in a previous study (Blom et al., 2019), it did not show to relate to the visual field in which the expression was shown. Hence, though the facial muscles might react to the facial expressions shown in the paradigm used in the present study, participants apparently can judge the emotionality of presented faces without relying on the sensorimotor route. Relatedly, the task utilized by Korb et al. (2016) might have relied more on the sensorimotor route than the current studies' task, hence providing a different account for their reported findings somewhat diverging from our present findings.
In closing, although the present study mainly aimed to address the role of facial functioning in emotional processing of facial expressions, we would like to stress that is of equal importance to study different facets of emotion processing in patients with a facial paresis, as well as in patients with cerebellar damage. Other studies have for example reported differences in emotion perception and regulation in individuals with cerebellar damage (e.g., Houston et al., 2018). We, therefore, believe that future studies could examine this further by use of additional tasks that have previously been proven insightful for individuals with facial paresis and/or cerebellar damage. We wish to note here that the current study is part of a larger project that examined possible differences in emotion processing of facial expressions as well as perceived quality of life, social function, and emotion between VS patients with and without facial paresis. This project aims to provide a first step in obtaining a more complete picture of emotion processing and emotion regulation in patients by using several experimental tasks as well as questionnaires (see  Aarts, Kunst, Wever & Semin, manuscript under review).
The current study is one of the few experimental studies on facial emotion processing in patients with a facial paresis, and patients with a VS in particular. Knowledge on emotion processes that are and that are not affected in VS patients' with and without facial paresis informs health practitioners regarding the care they could provide patients with respect to their wellbeing. Although the present study suggests that facial paresis is not associated with impaired lateralization of emotion processing, future studies could focus on other types of facial emotion processing to further the understanding of the possible impact of a facial paresis on emotion processing. Committee of the Leiden University Medical Center. Protocol number: NL40223.058.12, principal investigator C.C. Wever.

CO N FLI C T O F I NTE R E S T
The authors report no conflict of interest.

AUTH O R CO NTR I B UTI O N
SB and GS designed the study. SB programmed the study, collected the data, prepared, and analyzed the data. SB wrote the manuscript in consultation with HA and GS. CW and HK helped in access to participants, and provided consultation on the parts of the manuscript about the VS patient sample specifically.

DATA AVA I L A B I L I T Y S TAT E M E N T
A data set is available and stored and can be requested from the corresponding author.