Facial expression recognition is an important ability that provides crucial information about others’ internal states. The extent to which emotional facial expressions are differentially processed by areas within the right and left hemispheres of the brain is a topic of significant debate. Much of this debate has focused on three theories of hemisphere asymmetries in emotion processing. The Right Hemisphere Hypothesis (RHH; e.g., Borod et al., 1998) argues that the right hemisphere is dominant for the processing of all emotional stimuli and is therefore dominant for the processing of all emotional facial expressions. The Valence Specific Hypothesis (VSH; e.g., Adolphs et al., 2001), in contrast, proposes that the right hemisphere is dominant for the processing of negative emotions but the left hemisphere is dominant for the processing of positive emotions. According to this theory, negative facial expressions such as anger, disgust, fear, and sadness produce right hemisphere dominance, while positive facial expressions such as happiness and surprise produce left hemisphere dominance. A similar theory, the Approach/Withdraw Hypothesis (AWH; e.g., Harmon-Jones, 2004), argues that the right hemisphere is dominant for withdrawal emotions and the left hemisphere is dominant for approach emotions. The difference between this theory and the VSH is that it classifies anger as an approach (left hemisphere) emotion rather than a negative (right hemisphere) emotion.

Previous research on facial expression processing is mixed in its support of these theories. Some studies observe results that align with the RHH (Bourne, 2010; Ross et al., 2013; Innes et al., 2016; Damaskinou and Watling, 2018; Kajal et al., 2020); other studies observe results that are more in line with the VSH (Jansari et al., 2011; Kumar and Srinivasan, 2011; Prete et al., 2014a; Lichtenstein-Vidne et al., 2017); few studies, if any, provide direct support for the AWH. As such, a growing number of researchers have suggested that the truth may be some combination of the RHH and VSH perspectives (Killgore and Yurgelun-Todd, 2007; Rahman and Anchassi, 2012; Abbott et al., 2013; Najt et al., 2013; Prete et al., 2015a, b, 2018; Lai et al., 2020).

Critically, despite the important differences between the theories, the RHH, VSH, and AWH are similar in that they all predict right hemisphere dominance for the negative facial expressions of sadness, fear, and disgust. This effect is observed by many studies (e.g., Bourne, 2010; Kumar and Srinivasan, 2011; Rahman and Anchassi, 2012; Innes et al., 2016; Damaskinou and Watling, 2018; Kajal et al., 2020; Hausmann et al., 2021). Nonetheless, some studies have observed left hemisphere dominance for one or more of these expressions under particular experimental conditions (Prete et al., 2014b, 2015a; Burt and Hausmann, 2019; Bublatzky et al., 2020; Lai et al., 2020; Stanković and Nešić, 2020). As such, regions within the left hemisphere may play a greater role in processing facial expressions than predicted by the RHH, VSH, or AWH theories.

One factor that appears to play a role is the number of facial expressions included in the task. A study by Abbott et al. (2014) suggests that left hemisphere regions are more involved when the task requires recognizing a larger compared to a smaller number of expressions. Patients with right or left hemisphere damage and control subjects performed three blocks of an identification task in which they labeled emotional facial expressions. The number of expressions to be identified varied from two (happy and sad) in the first block, four (happy, sad, surprised, and disgusted) in the second block, and all six basic emotions (happy, sad, surprised, disgusted, angry, and fearful) in the third block. When two facial expressions were tested, patients with right hemisphere damage were impaired compared to patients with left hemisphere damage for both happy (positive) and sad (negative) expressions, in line with the RHH. When four or six facial expressions were tested, however, right and left hemisphere damage patients were equally impaired compared to controls for positive and negative emotions. Thus, areas within the left hemisphere contribute to facial expression processing to a greater extent when more facial expressions are included in the task, and therefore, number of expressions included could play a role in producing left hemisphere dominance in negative facial expression processing. Importantly, however, more of the studies observing left hemisphere dominance for negative expressions use two facial expressions (Prete et al., 2014b, 2015a; Burt and Hausmann, 2019; Bublatzky et al., 2020; Lai et al., 2020) than six (Stanković and Nešić, 2020). As such, other contributing factors must be considered.

Another factor is the particular task that subjects perform. Studies of hemisphere asymmetries in facial expression processing have employed a wide variety of tasks. Some studies ask subjects to decide which of two faces is more emotional (Bourne, 2010; Innes et al., 2016; Hausmann et al., 2021) or more like a particular emotion (Jansari et al., 2011; Rahman and Anchassi, 2012); others ask whether a face is emotional or not (Najt et al., 2013; Stanković and Nešić, 2020), a positive or negative expression (Lichtenstein-Vidne et al., 2017; Kajal et al., 2020), or the same or different expression compared to another (Mattavelli et al., 2013, 2016; Flack et al., 2015; Sliwinska and Pitcher, 2018). Of note, none of these tasks require subjects to explicitly identify different facial expressions (i.e., respond whether an expression is happy, angry, and sad, etc.), and studies that do require subjects to explicitly identify facial expressions are rare. Indeed, of studies that include all six emotional expressions, Abbott et al. (2014) and Ouerchefani et al. (2021) are the only ones in which subjects are asked to identify facial expressions in photographs. Both studies compare right and left hemisphere damage patients to control subjects and observe equivalent impairment for right and left hemisphere damage patients relative to controls for positive and negative expressions. In contrast, studies that include all six expressions and use tasks that do not require explicit identification (e.g., Bourne, 2010; Najt et al., 2013) observe right hemisphere dominance for most negative expressions.

Explicit identification of facial expressions in the context of multiple expressions may involve left hemisphere regions to a greater extent than tasks that do not require identification, due to the influence of linguistic processes that function dominantly in the left hemisphere. The influence of linguistic categories on left hemisphere facial expression processing was demonstrated in a study by Burt and Hausmann (2019) in which healthy subjects were sensitive to the linguistic category (happy vs. sad, angry vs. fearful) of facial expressions when they were presented to the left hemisphere but not when they were presented to the right. The researchers conclude that category-based processing of facial expressions is due to left hemisphere involvement. As such, tasks that require identifying the linguistic category of facial expressions may be more likely to involve left hemisphere regions than tasks that do not require this information.

The present research investigated this possibility in two experiments. Both experiments included all six basic emotional facial expressions presented to the right or left hemisphere using a divided-visual field paradigm. In Experiment 1, participants identified emotional expressions by pushing a key corresponding to one of six labels. In Experiment 2, participants detected emotional expressions by pushing a key corresponding to whether an expression was emotional or not. Left hemisphere dominance in negative expression processing was expected during Experiment 1, when an identification task was used, but not during Experiment 2, when a detection task was used.

Three major theories of hemisphere asymmetries in emotion processing—the RHH, VSH, and AWH—predict right hemisphere dominance for negative facial expressions of disgust, fear, and sadness. This is observed in many studies (e.g., Bourne, 2010; Kumar and Srinivasan, 2011; Rahman and Anchassi, 2012; Innes et al., 2016; Damaskinou and Watling, 2018; Kajal et al., 2020; Hausmann et al., 2021), however, some studies observe left hemisphere dominance for one or more of these expressions (Prete et al., 2014b, 2015a; Burt and Hausmann, 2019; Bublatzky et al., 2020; Lai et al., 2020; Stanković and Nešić, 2020). The present research investigated whether tasks that require identifying the linguistic category of facial expressions are more likely to produce left hemisphere dominance than tasks that do not require this information. In line with this idea, fearful facial expressions exhibited a left hemisphere advantage during an identification task (Experiment 1) but not during an emotion detection task (Experiment 2). In contrast to this idea, however, sad expressions exhibited a left hemisphere advantage during both identification and detection tasks, and happy facial expressions exhibited a left hemisphere advantage during the detection task but not during the identification task. As such, the present study provides partial support for the idea that identification increases left hemisphere involvement and raises additional questions.

One question is why fearful facial expressions were the only expressions that exhibited greater left hemisphere involvement in the identification task than the detection task, when it was predicted that all negative expressions, as well as positive expressions, would show this effect. Research suggests that highly arousing emotions such as fear are processed rapidly by the amygdala and related subcortical regions (Le Doux, 2007; Lindquist et al., 2012). Indeed, Fusar-Poli et al. (2009) observe greater bilateral amygdala activity for fearful expressions compared to other expressions in a meta-analysis of fMRI studies of facial expression processing. It is possible, therefore, that the detection of fear relies on subcortical structures that do not exhibit a hemisphere asymmetry while the identification of fear requires more elaborate processing by cortical systems that do exhibit a hemisphere asymmetry. This is similar to a model proposed by Shobe (2014) in which emotional stimuli are processed initially by subcortical regions, and secondarily by cortical regions in the right and left hemispheres that process the emotional information more fully. Shobe argues that subcortical regions are right lateralized, however, which is not in line with the lack of hemisphere asymmetry for fearful expressions in the detection task. In contrast to fear, less arousing emotions such as happy and sad may rely on cortical systems for both identification and detection. Indeed, the similar hemisphere asymmetry observed in the identification and detection tasks for sad facial expressions is in line with this idea. Nonetheless, the effect of task observed for happy facial expressions (left hemisphere advantage during detection but not identification) is not. Thus, this cannot be the complete explanation.

Although a similar hemisphere asymmetry in identification and detection tasks for sad facial expressions is in line with the idea that low-arousal emotions rely on the same cortical systems regardless of task, the left hemisphere advantage in both tasks contrasts with the hypothesis that left hemisphere involvement for negative facial expressions is due to the need for linguistic categorization, as this is not needed for emotion detection. Results from Lai et al. (2020) suggest another possible explanation. In this study, N170 latencies in the left hemisphere were shorter for sad than happy expressions when faces were presented for a brief amount of time (14 ms), implying that left hemisphere regions play a role in the early processing of sad facial expressions. If so, this could explain the left hemisphere advantage for sad facial expressions in the detection task.

Nonetheless, the question of why left hemisphere regions might be important for processing sad facial expressions remains. Sadness is characterized by the AWH as a “withdrawal” emotion that should evoke right hemisphere dominance, and indeed, as noted previously, there is much research to support this interpretation (e.g., Bourne, 2010; Kumar and Srinivasan, 2011; Najt et al., 2013; Innes et al., 2016). However, others have argued that sad facial expressions may evoke prosocial or caregiving responses in the viewer (Tooby and Cosmides, 1990; Lazarus, 1991). From this perspective, sadness could be reconceived as an “approach” emotion that should, according to the AWH, produce left hemisphere dominance. In line with the idea that sad facial expressions elicit a prosocial/caregiving response, Colasante et al. (2017) observed a larger N170 response in adult participants for sad infant faces than for sad adult faces or happy faces. Critically, however, the N170 response was in the right hemisphere rather than the left, contradicting the idea that sadness is left dominant. Clearly further research will be necessary to explain this effect.

In contrast to left hemisphere dominance for sad facial expressions, left hemisphere dominance for happy facial expressions is predicted by both the VSH and AWH, as happy expressions are both positive and approach-oriented. The left hemisphere advantage for happy facial expressions observed during the detection task is in line with this prediction. It is unclear, however, why this effect was not observed during the identification task, which was hypothesized to require greater left hemisphere involvement than the detection task due to its linguistic nature. Future research will be required to fully understand hemisphere asymmetries in happy facial expression processing. Nevertheless, the present research highlights the complexity of the issue and the need to consider task demands.

In addition to considering task demands, future research should include all six basic emotional expressions, as was done in the present experiments. This is important because it increases left hemisphere involvement compared to only including two expressions (Abbott et al., 2014). Moreover, including all six basic emotional expressions in future research will increase the comparability of studies by ensuring that the same emotional expressions are tested. Currently, many studies only include two emotional facial expressions (Killgore and Yurgelun-Todd, 2007; Kumar and Srinivasan, 2011; Mattavelli et al., 2013, 2016; Prete et al., 2014a, 2015a, b, 2018; Colasante et al., 2017; Lichtenstein-Vidne et al., 2017; Damaskinou and Watling, 2018; Bublatzky et al., 2020; Kajal et al., 2020; Lai et al., 2020) and which two are included varies [e.g., sad (Damaskinou and Watling, 2018; Lai et al., 2020) vs. angry (Prete et al., 2014a, 2015a, b, 2018; Lichtenstein-Vidne et al., 2017)], making studies difficult to compare. In addition, studies including only two expressions rarely include expressions of disgust or surprise, and therefore, these expressions have not received as much attention as others. Neither disgust nor surprise exhibited a hemisphere asymmetry in the present experiments, but it will be crucial to know more about these expressions for a complete understanding of asymmetries in expression processing. Finally, future facial expression research should include all six expressions because tasks that require processing of six expressions are closer to real-life facial expression processing, which includes a wide variety of expressions, than tasks that only require two, and as such, provide a more realistic perspective on the systems and processes involved.

Future research will also benefit from the use of different methods for assessing hemisphere asymmetries in expression processing. A recent ERP study observed widespread bilateral activity during divided-visual-field presentations of facial expressions (Prete et al., 2018), calling into question the extent to which divided-field presentations preferentially activate left or right hemispheres. Moreover, this same study observed different support for the RHH and VSH depending on the measure examined, with behavioral performance supporting the VSH and ERP results supporting the RHH. As such, it cannot be assumed that neuroimaging methods will produce similar results as the present study. Other research indicates that whether an expression is perceived consciously or unconsciously affects the pattern of hemisphere asymmetries, and suggests that unconsciously perceived expressions may preferentially activate the left or right hemispheres more effectively than consciously perceived expressions (Prete et al., 2015a). Expressions in the present study were consciously perceived; thus future research may benefit from presentations in which the expression is perceived unconsciously.

In conclusion, the present research adds to a growing body of work on facial expression processing that does not align with major theories of hemisphere asymmetries in emotion processing (Killgore and Yurgelun-Todd, 2007; Tamietto et al., 2007; Rahman and Anchassi, 2012; Abbott et al., 2013; Najt et al., 2013; Prete et al., 2015a, b, 2018, 2019; Lai et al., 2020). Clear evidence for left hemisphere involvement in processing the negative facial expressions of fear and sadness was observed, which contrasts with predictions of previous theories and provides a foundation for further research into the nature of this role. The present research also provides clear evidence for the influence of task demands on hemisphere asymmetries for certain expressions, namely fearful and happy. Thus, task demands should be considered in future investigations.

The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Psychology Review Board at Macalester College. The patients/participants provided their written informed consent to participate in this study.

EB was responsible for the conception and design of the experiments, the statistical analyses, and wrote the manuscript.

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.