Comparing two versions of the Chimeric Face Test: A pilot investigation

ABSTRACT

The Chimeric Face Test (CFT) is a widely used behavioral measure of degree of lateralization for emotion processing. The Pictures of Facial Affect (Ekman, 1976 [Pictures of facial affect. Consulting Psychologists Press.]) have often been used to create chimeras for this task but have widely been critiqued due to lack of ethnic diversity and small stimuli numbers. In this brief study participants (N = 45) completed two Chimeric Face Tests, one using the Pictures of Facial Affect and one using the NimStim facial stimuli (Tottenham et al., 2009 [The NimStim set of facial expressions: Judgments from untrained research participants. Psychiatry Research, 168(3), 242–249]). The laterality scores were compared across measures. The results show the two measures are related; laterality quotients showed a strong correlation between the two tasks. Participants showed a left-visual field bias on both tasks, indicative of a right-hemisphere bias for the processing of emotions. The NimStim Chimeric Face Test however was found to give a more conservative estimate of degree of lateralization. The NimStim Chimeric Face Test is discussed as a valid measure for examining lateralization for emotion processing, The extent to which different versions of the Chimeric Face Test are comparable is discussed.

KEYWORDS:

• Lateralization for emotion processing
• Chimeric Face Test
• right-hemisphere
• laterality
• left-visual field bias

Researchers have argued that similar to other cognitive functions, such as language (Groen et al., 2013) and spatial ability (Vogel et al., 2003), the two hemispheres of the brain play independent roles in the processing of emotions (Davidson, 1992; Killgore & Yurgelun-Todd, 2007). Three prominent theories have been put forward to explain how emotions are processed in the brain: the valence hypothesis (Davidson, 1992), the approach-withdraw (Demaree et al., 2005) hypothesis and the right hemisphere hypothesis (Borod, 1992; Smith & Bulman-Fleming, 2005; Watling et al., 2012). The valence hypothesis posits that emotional stimuli with a positive valence are processed in the left hemisphere, and emotional stimuli negative in valence are predominantly processed in the right hemisphere, whereas the approach-withdrawal theory suggests that emotions associated with approach behaviours are processed in the left hemisphere and emotions associated with withdrawal behaviours are processed in the right-hemisphere. Whilst these theories have some support (see Adolphs, 2001; Borod, 1992; Davidson et al., 1990), the right-hemisphere hypothesis is often cited as the dominant theory of lateralisation of emotion processing. The right-hemisphere hypothesis states that emotional stimuli are perceived more efficiently by the right-hemisphere than by the left-hemisphere, irrespective of the valence of the emotional stimuli (Borod, 1992; Smith & Bulman-Fleming, 2005; Watling et al., 2012).

There has been ample research in support of a right hemisphere dominancy in the processing of emotions in both child (Bava et al., 2005; Workman et al., 2006) and adult populations (Bourne, 2005; Nakamura et al., 1999; Spence et al., 1996). Further, lesion studies have provided convincing evidence in support of the right hemisphere hypothesis of emotion processing (Adolphs et al., 1996; Borod, 1992). For example, Borod (1992) carried out a lesion analysis on participants with focal brain damage and found that facial emotion recognition performance was negatively correlated with damage to the right hemisphere regions. Importantly, patients with lesions in the left hemisphere did not show impairments in their facial emotion recognition. These findings support that cortical systems of the right hemisphere are important in the processing of facial emotions. Similar patterns have been found using event related potentials when viewing emotional expressions, with greater activation in the right-hemisphere compared to the left-hemisphere in children (Batty & Taylor, 2006).

Dating back to the early 1930s (Wolff, 1933) and used extensively ever since, the Chimeric Face Test (CFT) is a well-established free-viewing behavioural task that has been widely used as a measure of hemispheric lateralisation for emotion processing, and the extent to which different emotions are lateralised to either hemisphere of the brain (Bourne, 2008; Levine & Levy, 1986; Levy et al., 1972; Workman et al., 2006). A chimeric face is created by splicing an emotive and neutral face down the centre and then reconfiguring to merge one emotive half face with one neutral half face to create a chimeric image. The mirror image is taken, so there are two chimeras. Typically, one chimera image is placed on the top and the second chimera on the bottom (mirror images of each other). Within the task, faces with the emotive half on the left appear on top in half the trials (with faces with the emotive half on the right appearing below), and in half the opposite is true. In the CFT tasks, participants are typically asked which of the two chimeras looks more emotive. Tasks assessing hemispheric lateralisation for emotion processing in the brain typically take advantage of the cross organisation of the visual pathways in the brain, whereby information viewed in an individual’s left visual field is primarily processed in the right hemisphere of the brain and information presented in the right visual field is primarily processed in the left hemisphere of the brain (Bourne, 2008, 2010). As such, on a CFT, if the individual selects the face with the emotion on the left-side of the image (left visual field) as more emotive then this is believed to reflect right hemisphere processing, whereas if the individual selects the face with the emotion on the right side of the image (right visual field) as more emotive then this is believed to reflect left hemisphere processing (Beaumont, 1983).

There is a considerable amount of evidence to support the CFT as a valid measure of lateralisation for emotion processing (Bava et al., 2005; Bourne, 2006, 2010; Coronel & Federmeier, 2014; Damaskinou & Watling, 2018; Killgore & Yurgelun-Todd, 2007; Kucharska-Pietura et al., 2003). For example, evidence with unilateral brain damage patients has shown that patients with left hemisphere brain damage show a left visual field bias on CFTs (indicating right hemisphere dominancy), whereas those with unilateral right hemisphere brain damage show reduced left visual field bias (indicating reduced right hemisphere processing; Kucharska-Pietura et al., 2003). Similar patterns have also been demonstrated in children with unilateral brain damage (Bava et al., 2005). The CFT has also been validated with EEG, Damaskinou and Watling (2018) demonstrated that adults showed patterns of greater amplitude within the right hemisphere than left hemisphere when chimeras displayed the emotion in the left visual field; in contrast, chimeras with the emotion displayed in the right visual field showed patterns of greater activation within the left hemisphere compared to the right hemisphere.

Current limitations of the Chimeric Face Test (CFT)

The Pictures of Facial Affect (Ekman, 1976) have been widely used to assess facial emotion recognition and dominate much of the literature (e.g., Ekman & Friesen, 1971; Lawrence et al., 2015; Thomas et al., 2007; Workman et al., 2006) and in research employing Chimeric Face Tests (CFTs). Despite this, the Pictures of Facial Affect have increasingly received criticism due to the lack of ethnic and racial diversity in the set (Phillips et al., 1998) and containing too few stimuli (Winston et al., 2013). To address these concerns Tottenham and colleagues (2009) developed the NimStim facial stimulus set (Tottenham et al., 2009). The NimStim facial stimulus set contains 672 coloured photographs from 43 (25 male) professional actors posing eight emotional expressions (the six basic emotions, plus neutral and calm). This stimulus set is available in colour and uses a range of ethnic and racial diversities. In the development of the NimStim facial stimulus set, Thomas and colleagues (2009) provided evaluation of both the reliability and validity of this stimulus set. Validity was tested by asking participants to view and label all 672 photographs using a forced choice paradigm (participants chose from six basic emotions, calm, neutral and none of the above) and the proportion of individuals who correctly labelled the intended expression were recorded. In order to account for agreement that may have occurred by chance, Cohens kappa (Cohen, 1960) was calculated. After a short break, participants were asked to repeat the labelling task, and the proportion of agreement between the two sets of ratings was calculated as a measure of reliability. In all instances the values obtained for these measures ranged from 0–1 and can be found for each model and emotional expression in supplementary materials provided by Tottenham and colleagues (2009).

In summary, researchers have consistently shown overall patterns of left-visual field bias on CFTs, believed to reflect right-hemisphere processing. As well as overall patterns of right-hemisphere processing, these findings extend to the processing of individual emotions (e.g., Bourne, 2010; Workman et al., 2006). For example, Bourne (2010) found that all emotions showed significant lateralization to the right-hemisphere processing. Rsearchers are increasingly favouring developing their own Chimeric Face Tests (e.g., Blom et al., 2020; Gupta & Pandey, 2010), but to date, there has been little consideration or comparisons of different versions of the Chimeric Face Test. The aims of this report are to examine the extent to which two Chimeric Face Tests are comparable. In order to demonstrate the concurrent validity of the CFT using the NimStim stimuli as a valid measure of lateralization of emotion processing, participants in this study will complete both the well-utilized Chimeric Face Test by Workman et al. (2006). using the Pictures of Facial Affect stimuli set and a new Chimeric Face Test created using the NimStim facial stimuli (Tottenham et al., 2009), Chimeras will be presented for each of the six basic emotions (i.e., happy, sad, angry, fear, surprise, disgust). Internal reliability will be assessed by examining the inter-correlations in the emotion laterality quotients obtained from both tasks. Given that the CFT should provide general patterns of lateralization for emotion processing, it is expected that participants’ patterns of lateralization on the two tasks will be similar.

Method

Participants

The sample consisted of 50 participants (M_age = 22.12, SD = 3.26; range 18–32, males = 15), recruited from Royal Holloway, University of London. Participants took part for course credit. Participants were asked to self-report handedness (left, right, ambidextrous), strength of handedness was further assessed through a handedness measure (Dorthe et al., 1995), where participants responded to 14 statements, indicating the extent to which they used their left or right hand for a variety of activities. Participants responded on a 7-point Likert scale that ranged from “Always with Left” (−3) to “Always with Right” (+3). Responses were totaled to get a score from - 42 to + 42 whereby a score of 0 would be indicative of no dominancy (equal use of left of right hand), a negative score would indicate left-hand dominancy and a positive score would be indicative of right-hand dominancy. Five participants identified as being left-handed; this was supported by the handedness measure (M = - 13.40). Consistent with past research, individuals who identified as being left-handed were removed from further analyses due to reported differences in patterns of lateralization for left and right handers (Bourne, 2008; Burton & Levy, 1989; Hellige et al., 1994). The mean handedness score for participants who identified as right-handed was 32.96 (SD = 7.47). The final sample consisted of 45 participants (M_age = 22.44, SD = 3.27, range 18–32; males = 12). A post-hoc sensitivity analysis was conducted using G*Power 3. Based on the sample of 45, and power of .80, we are able to detect an effect of .36 (moderate effect).

Materials and measures

Participants completed a measure of handedness (Dorthe et al., 1995), reported above and two computerized Chimeric Face Tests (CFTs), one using the Pictures of Facial Affect stimuli and one developed using the NimStim facial set (see Figures 1 and 2). The two CFTs were programmed using E-Prime 2.0, which was used for stimuli presentation. Participants took part on a laptop and made responses using the keyboard.

Figure 1. Examples of the Pictures of Facial Affect chimeras, used in previous research (e.g., Workman et al., 2006). From left to right faces represent anger, sadness, fear, disgust, happiness and surprise.

Figure 2. Examples of the NimStim chimeras created. From left to right faces represent anger, sadness, fear, disgust, happiness and surprise.

The NimStim Chimeric Face Test (CFT)

As aforementioned, research using the Chimeric Face Test has predominantly used images created from the Pictures of Facial Affect (Bourne & Maxwell, 2010; Bourne & Vladeanu, 2013; Drebing et al., 1997; Ekman, 1976; Innes et al., 2016; Watling & Bourne, 2013; Workman et al., 2006), typically containing one male and one female model (see Workman et al., 2000 for development of Pictures of Facial Affect CFT). In developing the NimStim CFT, instead of one male and female used in previous CFTs, two males and two females for each of the six basic emotions, further given critiques of the Pictures of Facial Affect, a range of racial and ethnic diversities were included in this task.

Stimuli selection

Two male and female models were selected for each of the six basic emotions, that met criteria for based on the following criteria: (1) the reliability, validity and Cohen’s kappa all were greater than .80 (2) models must have a neutral face that also met this criterion (3) a range of racial and ethnic diversity. In total 14 models were used (six male, eight female) and 16 images (two male, two female for each emotion) were selected (Table 1).

Table 1. Model selection for emotion for the development of the NimStim CFT.

	Happy	Sad	Anger	Fear	Disgust	Surprise
Face 1	1^+	1^+	3¨	10^+	16^•	7^+
Face 2	14*	3¨	17^•	14*	19^•	14*
Face 3	20^+	20^+	20^+	36^+	20^+	36^+
Face 4	33^+	27^+	34^+	43*	36^+	43*

Note: Face 1 and 2 female stimuli, Face 3 and 4 male stimuli. Ethnicities: ^+­ Caucasian, * African-American, ^•Asian-American, ¨Latino-American.

Creating chimeric faces

A similar procedure to early work in this area was used to create the stimuli (see Levine & Levy, 1986; Workman et al., 2006). For each image created, a model 100% intensity emotion expression image and the same models’ neutral face image were vertically split in half at the nose; this yielded 4 facial half images. Using the facial half images, two chimera images were created by taking each half of an emotive facial image (left and right side) and splicing it with the opposite half of the neutral facial image (i.e., right and left side, respectively) to create a “full face” (the nose was used as the main matching point).

For each of the two chimera images, a mirror image was created so to have an identical image, with the emotion and neutral sides swapped (i.e., where the image had the emotion on the left side of the face and neutral on the right, the mirror image had the emotion on the right side of the face and neutral on the left). Using Adobe Photoshop (CC 2015) an oval mask was placed around the face to remove hair and ensuring faces were of similar shapes and sizes. Further, to ensure consistency with the previous Pictures of Facial Affect Chimeric Face Test, chimeras were saved as greyscale. For each emotion, there were four chimeric images in total per model, for a total of 16 chimeras per emotion. In total, for the six basic emotions, there were 96 chimeras created.

Pictures of Facial Affect Chimeric Face Test. The Pictures of Facial Affect Chimeric Face Test (CFT), was created by Workman and colleagues (2006). As highlighted earlier the CFT stimuli were created using one male and female model for each of the six basic emotions. These stimuli were created in a similar way to the NimStim CFT created in this study. All images were shown in greyscale. In total there were eight images per emotion (four original chimeras and a mirror image for each; 48 images in total), each with the emotion presented either on the left or right of the chimera face, with neutral emotion on the other half of the face. Trial presentation included two chimeras (an original and the mirror image), one on the top and one on the bottom (see Figure 1 for example). There were eight trials per emotion as per the original task (four with the original chimera on the top and mirror image on the bottom, and four with the mirror image on the top and original chimera on the bottom). Trials were blocked by emotion (six blocks in total) and were randomized within each block. Cronbach’s alpha indicated a good level of internal consistency on this task (α = .85).

NimStim Chimeric Face Test. The task is identical to the Pictures of Facial Affect Chimeric Face Test (CFT) task with two exceptions (1) the use of the NimStim chimeric images (2) given there are four models for each emotion, allowing for the creation of eight original chimeras, there are 16 trials per emotion (96 trials in total). Cronbach’s alpha indicated an excellent level of internal consistency on this task (α = .90).

Procedure

Participants were seated individually in a quiet room. Following reading the information sheet, and having the opportunity to ask questions, they completed a consent form followed by the handedness measure using pen and paper. Next, participants completed both Chimeric Face Test (CFT) tasks on a 15-inch Lenova laptop at a viewing distance of approximately 50 cm. E-Prime 2.0 was used for presentation of both CFTs. The order of the two CFT tasks were counterbalanced; half of the participants started with the Pictures of Facial Affect CFT (Workman et al., 2006) and half of participants started with the NimStim CFT.

The emotion blocks within each task were randomised, as well as the trials within each emotion block through E Prime 2.0. Within each emotion block, participants were first instructed to read the block instructions before the trials begun. Participants were instructed to “decide which of the two faces look [happier, sadder, angrier, more scared, more disgusted, more surprised]”. In each trial, the instructions were followed by the automatic presentation of the mirrored chimera images (presented centrally on light grey background with one above the other). Participants responded using the up and down arrow corresponding to the face they believed looked more emotive (up arrow indicating the top face was more emotive, down arrow indicating the bottom face was more emotive). This was a free-viewing task whereby faces remained on screen until a decision was made, but participants were instructed to decide as quickly as possible. The two tasks together took 20 min to complete. Participants were fully debriefed on completion of the task as to the purpose of the study.

Scoring. Responses were recorded and summed separately for the number of times participants responded that the image with the emotion presented on the left side (left-visual field) was more emotive (indicative of a right hemisphere processing) and for the amount of times participants responded that the image with the emotion presented on the right side (right-visual field) was more emotive (indicative of a left hemisphere processing). To calculate a laterality quotient for each participant, we used the formula from Bourne (2008), see (Equation 1). This resulted in a calculated score for each emotion block from −1 to +1, whereby −1 indicated a left hemisphere processing bias, 0 indicated no-hemispheric bias, and +1 indicated a right hemisphere processing bias. An overall laterality quotient was calculated for each of the two CFT stimulus sets by averaging the scores of the six emotion blocks.

Equation 1. Equation used to calculate Laterality Quotient from Bourne (2008), for each emotion for each of the two CFTs separately. ${\displaystyle \frac{(\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{L}\mathrm{V}\mathrm{F}\mathrm{c}\mathrm{h}\mathrm{o}\mathrm{i}\mathrm{c}\mathrm{e}\mathrm{s}-(\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{s}-\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{L}\mathrm{V}\mathrm{F}\mathrm{c}\mathrm{h}\mathrm{o}\mathrm{i}\mathrm{c}\mathrm{e}\mathrm{s}\left)\right)}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{l}\mathrm{s}}}$

Results

The mean laterality quotient for the Pictures of Facial Affect Chimeric Face Test (CFT) was .28 (range = -.38 to + .88; SD = 0.34) and for the NimStim CFT mean = .18 (range = -.48 to + .92; SD = 0.35). Two one-sample t-tests were carried out to assess whether the overall laterality quotients for each stimulus set significantly differed from 0 (indicating no hemispheric dominance). The t-tests revealed that for both the Pictures of Facial Affect chimeras, t (44) = 5.58, p < .001, and the NimStim chimeras, t (44) = 3.46, p = .001, participants showed an overall left visual field bias (indicative of a right hemisphere processing). The overall laterality quotients from the two tests were strongly correlated, r = .84, p < .001. Given the NimStim CFT contained twice as many trials than the Picture of Facial Affect CFT, split-half reliability for the NimStim CFT was assessed by comparing the laterality quotient for all odd trials (8 trials per emotion) with all even trials. Split-half reliability showed excellent reliability, r = .93, p < .001.

Histograms were created to examine the distribution of overall laterality scores across the two tasks (see Figures 3 and 4). The distribution showed that for the Pictures of Facial Affect task was negatively skewed, with a higher proportion of participants showing a left-visual field bias (indicative or right-hemisphere processing bias). For the NimStim Chimeric Face Test the distribution of laterality quotients was normally distributed.

Figure 3. Histogram depicting percentage of participants and their corresponding overall laterality quotient for the Pictures of Facial Affect Chimeric Face Test. Dotted line indicates a score of 0 (no bias).

Figure 4. Histogram depicting percentage of participants and their corresponding overall laterality quotient for the NimStim Chimeric Face Test. Dotted line indicates a score of 0 (no bias).

A second set of one sample t-tests were completed for each emotion within each CFT task to assess if there was a laterality bias (significantly greater or lesser than 0 – indicating no bias). In line with typical protocol in this area (Bourne et al., 2009; Damaskinou & Watling, 2018), a Bonferroni correction was not applied when comparing each emotion laterality to zero. For the Pictures of Facial Affect CFT, all emotions showed significant left visual field bias and were significantly different from 0, ps ≤ .010, indicating a right hemisphere bias. For the NimStim CFT, all laterality quotients apart from anger (p = .082) were significantly different from 0, ps < .050, indicating a right hemisphere bias. Means and standard deviations for each task by emotion type are presented in (Table 2).

Table 2. Mean laterality quotients (range −1 to +1) for each emotion block for the Pictures of Facial Affect and the NimStim CFT. T-tests show significant differences between the two tasks.

	Pictures of Facial Affect	NimStim	t	p
Happy	.30 (.50)^+	.19 (.46)^+	1.82	.076
Sad	.13 (.32)^+	.18 (.37)+	0.64	.526
Anger	.39 (.39)^+	.12(.46)	−4.98	<.001**
Fear	.35 (.45)^+	.22 (.44)^+	−2.20	.033
Surprised	.33 (.50)^+	.15 (.46)^+	2.81	.007*
Disgust	.20 (.50)^+	.25 (.39)^+	0.75	.459

Note: **p ≤ .001; *p ≤ .008. Bonferroni correction applied α = .008. ⁺Indicates a laterality quotient significantly greater than 0 (right hemisphere processing).

Laterality quotients obtained for each emotion from both stimuli sets were compared to one another with separate paired sampled t-tests, with Bonferroni corrections applied to control for multiple comparisons (α = .008). For anger and surprise, but no other emotion, there was a significant difference in the laterality quotients obtained from the two CFT tasks, in both instances the laterality quotient from the Pictures of Facial Affect CFT was significantly higher than the NimStim CFT (see Table 2).

Inter-correlations

Inter-correlations between each emotion laterality quotient within each CFT task were examined to understand how strength of lateralization on one emotion might relate to strength of lateralization on another, as well as to the overall laterality quotient obtained from that CFT (see Tables 3 and 4). Given that the CFT is measuring emotion processing, it would be expected that the laterality quotients for each emotion should be correlated. The Pictures of Facial Affect CFT correlations between emotion laterality quotients were primarily moderate – strong (moderate, r = .30 to .50, strong r > .50); although, some correlations with the sad laterality quotient were not significant ps > .050. Individual emotion laterality quotients were strongly correlated with the overall laterality quotient, apart from sad, which was moderately correlated. For the NimStim CFT all correlations between emotion laterality quotients were moderately – strongly correlated (see Table 3; ps < .001 for all comparisons). Individual emotion laterality quotients were all strongly correlated with the overall laterality quotient (rs > .70).

Table 3. Inter-correlations between pictures of facial affect CFT stimulus emotions.

	Sad	Angry	Fear	Surprised	Disgust	Overall LQ
Happy	.34*	.51***	.66***	.60***	.61***	.84***
Sad		.23	.40**	.18	.18	.46***
Angry			.49**	.46***	.59***	.72***
Fear				.70***	.55***	.84***
Surprised					.62**	.81***
Disgust						.81***

Note: ***p ≤ .001; **p ≤ .050.

Table 4. Inter-correlations between NimStim CFT stimulus emotions.

	Sad	Angry	Fear	Surprised	Disgust	Overall LQ
Happy	.68***	.76***	.70***	.55***	.62***	.87***
Sad		.61***	.65***	.40**	.54***	.77***
Angry			.58***	.72***	.69***	.89***
Fear				.63***	.44**	.80****
Surprised					.58***	.80***
Disgust						.78**

Note: ***p ≤ .001; **p ≤ .050.

Correlations across-tasks

Correlations were examined for each emotion laterality quotient within the CFT across the two tasks, to understand to what extent strength of lateralization on one task might relate to strength of lateralization on the other task (see Table 5). Given the CFT is measuring emotion processing, it would be expected that the laterality quotients for each emotion across tasks should be correlated. Overall, correlations between the two task for each emotion were moderate to strong, with the exception of sad which did not correlate across the two tasks.

Table 5. Correlations for each emotion laterality quotient across tasks.

	Happy	Sad	Angry	Fear	Surprised	Disgust
Happy	.67**	.14	.57**	.64**	.69**	.59**
Sad	.62**	.21	.42*	.41*	.48**	.50**
Angry	.63**	.27	.66**	.66**	.76**	.72**
Fear	.55**	.12	.45**	.49**	.65**	.46**
Surprised	.47**	.25	.46**	.60**	.60**	.43*
Disgust	.52**	.22	.51**	.71**	.57**	.51**

Note: ***p ≤ .001; **p ≤ .050. Left column shows NimStim emotion laterality quotients, rows indicate Pictures of Facial Affect laterality quotients. Bold shows correlations between same-emotion quotients.

Discussion

The primary aim of this study was to validate a new Chimeric Face Test (CFT) to assess the lateralization for emotion processing using the NimStim facial stimulus (Tottenham et al., 2009). Participants’ patterns of hemispheric lateralization for facial emotion processing on the newly developed CFT were compared with a previously established and widely used CFT that had used the Pictures of Facial Affect stimuli to establish the validity of the newly established measure. Consistent with previous research in the area, there is evidence of an overall left visual field bias (right hemisphere) in the processing of emotions for both of the CFT measures. These findings are consistent with those widely documented in the literature (Watling et al., 2012; Workman et al., 2000). As well as this, the results show that the laterality quotients obtained from the NimStim CFT are strongly correlated to responses on a previously well-used CFT, using the Pictures of Facial Affect. These findings highlight that both tasks appear to be measuring the same thing.

Similar to previous work (e.g., Bourne, 2010), for the Pictures of Facial Affect CFT, it was found that for all emotions examined, participants tended to show evidence of right hemisphere processing (laterality quotients significantly different from 0). The findings in this study support the findings of previous work that demonstrates a right hemisphere processing of emotions (Bourne, 2010; Levine & Levy, 1986; Watling et al., 2012; Workman et al., 2006). For the newly developed NimStim CFT, the findings were similar to those with the Pictures of Facial Affect: laterality quotients obtained from each emotion were significantly different from 0 (indicating right hemisphere processing). However, there was one difference in that for anger only, the laterality quotient scores did not significantly differ from 0, although this was approaching significance (p = .082). In general, these findings indicate that the newly designed CFT with the NimStim stimuli set has a consistent pattern of findings as when using the Pictures of Facial Affect stimuli set.

Importantly, it was expected that there would be relationships between laterality quotients for facial emotion processing. When examining the inter-correlation between the laterality quotients for each emotion, all emotions (including anger) significantly and positively correlated with the overall laterality quotient for both the Pictures of Facial Affect CFT and the NimStim CFT. When examining the inter-correlations between laterality quotients for the different emotions, for the Pictures of Facial Affect CFT it was found that laterality quotients obtained for sad expressions did not significantly correlate with anger, surprise and disgust laterality quotients. In fact, when looking at the relationship with the overall laterality quotient, sad was the only emotion that showed moderate instead of strong correlations. Further, when correlating emotion laterality quotient across tasks, sad was the only emotion across the two tasks that did not correlate, with all other emotion comparisons showing strong correlations. For the NimStim CFT, all laterality quotients for each emotion correlated with all other emotion laterality quotients. These findings indicate that laterality quotients of each emotion for the NimStim CFT are more strongly related to one another than the Pictures of Facial Affect CFT.

When comparing laterality quotients obtained on both CFTs, there were some differences that emerged; for instance, the new NimStim CFT typically showed less hemispheric dominance than the Pictures of Facial Affect CFT (although all apart from anger showed significant right hemisphere dominancy). These differences were most pronounced for anger and surprise laterality quotients, with the Pictures of Facial Affect CFT laterality scores significantly higher (i.e., more right hemisphere dominant). It may be that scores on the NimStim CFT may provide a more conservative estimate (less strongly lateralised to the right hemisphere) due to the added racial diversity of the models included in trials. In fact, there is evidence that individual’s show more pronounced hemispheric asymmetry for own raced faces (Correll et al., 2011). Whilst data on ethnicity was not collected from the sample, regardless of participant ethnicity the mix of racial and ethnic diversity of the NimStim CFT may affect emotion judgements and is likely to result in differences in laterality quotient from the two measures. In contrast, the Pictures of Facial Affect contained only Caucasian models. According to the findings from Correll et al. (2011), it would be expected that using a combination of Caucasian and other ethnic stimuli would result in a lower overall laterality quotient score for all participants. Although beyond the scope of this study, future research is needed to examine how participant ethnicity may impact laterality quotient obtained for racially diverse stimuli.

Limitations of the current work should be acknowledged, primarily the small sample size used within this study. Whilst the study was powered to detect an effect of .36, the ability to detect smaller effects present is reliant on the need for a larger sample. This work however provides preliminary evidence that using the conclusions drawn from studies using the Pictures of Facial Affect CFT should be similar to those that make use of the NimStim stimuli for a CFT, especially when assessing relationships. Albeit, it should be noted that the NimStim CFT provides a more ecologically valid assessment of degree of lateralisation given the increased diversity within the stimuli set. In fact, this may raise questions as to what extent previous CFT may have been over or under-estimating degree of lateralization depending on their sample characteristics (i.e., ethnicity), where raw laterality quotients were of interest (opposed to assessing relationships between variables).

Conclusion

In conclusion, this study finds evidence that the newly developed NimStim CFT is a valid measure of laterality for emotion processing and is comparable to the Pictures of Facial Affect CFT used in previous research. Findings support that there is a relationship between the strength of lateralization for the six emotions and that the scores obtained from the NimStim CFT are highly correlated with the Pictures of Facial Affect CFT, which has been widely used within the literature. Importantly, it should be noted that scores on the newly developed NimStim measure may be more conservative estimate of hemisphere dominance, given the inclusion of racial diversity, arguably a more ecologically valid measure of emotion processing. In the past few years, researchers have begun to publish work using a CFT developed with stimuli from the NimStim stimulus set (e.g., Damaskinou & Watling, 2018; Watling & Damaskinou, 2020) this short report, may provide insight to researchers as to what extent different versions of the CFT are comparable.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The participants of this study did not give written consent for their data to be shared publicly.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.