Abstract
A recent meta-analysis of functional neuroimaging contrasts between emotional and neutral face processing has shown that the processing of facial emotions can be better classified according to threat detection than emotional valence, with the authors suggesting that their data are inconsistent with both the right-hemisphere and valence models of emotional laterality. I report empirical and theoretical data indicating that facial expressions are better classified according to threat detection than to the distinction between positive and negative emotions. I challenge, however, the claim that laterality effects provide little support to the right-hemisphere model of emotional laterality. This claim contrasts with neuropsychological and psychophysiological investigations that have shown that the right hemisphere has a graded prevalence for emotions provoked by threatening events. A reanalysis of data obtained in the target study suggests that the reported data are not necessarily inconsistent with a model assuming a graded, right-hemisphere dominance for emotions. I present a model of hemispheric asymmetries that could be consistent with the assumption that the right hemisphere’s dominance for emotions may mainly be concerned with threatening events.
Introduction
In a recent article, Lukito and collleagues1 published an updated meta-analysis of functional neuroimaging contrasts between emotional (e.g., angry, happy) and neutral face processing, taking into account the problem of emotional lateralization.
These authors showed that the processing of facial emotions can be better classified according to threat detection than to emotional valence and that laterality effects provide little support to either the right-hemisphere or the valence models of emotional laterality. I consider these 2 different aspects of their conclusions.
The first claim, concerning the classification of facial emotions, was based on the fact that, relative to neutral faces, only fearful and angry expressions provoked extensive activation clusters in different cortical and subcortical structures, whereas happy and sad faces did not evoke significant activation. Furthermore, clustering analyses showed similar activation patterns of fearful and angry faces, whereas happy and sad expressions had the lowest correlation with and highest distance from all the other emotional faces. Fearful faces evoked the most extensive clusters of activation in the brain, which was considered consistent with how fearful faces increase alertness for potential threats2 and heighten the levels of vigilance and arousal that are needed to promote the readiness for fight-or-flight responses.3
The second claim, concerning emotional lateralization, was based on how activation clusters provoked by fearful and threatening faces were observed in both right and left cortical and subcortical structures, providing little support to right-hemisphere dominance or the valence model of emotional laterality.
In my opinion, Lukito and colleagues’ statement that processing of facial expressions can be better classified according to threat detection than to the distinction between positive and negative emotions is supported not only by results of the their meta-analysis, but also by other empirical and theoretical reasons. Less convincing, however, is their statement that laterality effects provide little support to the right-hemisphere model of emotional laterality.
Ekman’s model of the basic emotions,4 which underlies the classification of facial emotions, is only partly consistent with the distinction between positive and negative emotions. Not all positive and negative emotions equally satisfy all the characteristics that should distinguish basic emotions, and emotional expressions that best satisfy these criteria are those dealing with threatening events. In particular, Ekman noted that a distinctive universal sign (to distinguish a basic emotion from others, across cultures) is present in all the basic negative emotions but not in certain positive emotions. He attributed this difference to the survival value of distinctive signs in the case of negative (but not of positive) emotions. Furthermore, other characteristics that distinguish basic emotions (such as quick onset, automatic appraisal, unbidden occurrence, and distinctive physiology) apply more to threat detection than to other categories of basic emotions. For these reasons, I quite agree with the Lukito and colleagues’ statement that the processing of facial emotions can be better classified according to threat detection than to the distinction between positive and negative emotions.
I do not equally agree, however, with their claim that their data are at odds with the model assuming right-hemisphere dominance for emotions, given that other lines of research have supported the right-hemisphere hypothesis, a different analysis of the data obtained by Lukito and colleagues supports the model, and a model of hemispheric asymmetries implies that right-hemisphere dominance for emotions may mainly concern life-threatening events.
Research supporting the right-hemisphere model of emotional laterality
Two main lines of research have suggested that the right hemisphere may have a graded prevalence for emotions (i.e., a prevalence greater for some emotions than for other emotions) and that this prevalence may mainly concern emotions provoked by threatening events. Neuropsychological studies of comprehension and expression of emotions at the facial or vocal level among healthy participants and patients with unilateral brain lesions have indicated a graded, right-hemisphere prevalence that favours negative emotions more than positive emotions. A review of studies on this subject can be found in Table 1 of a recent survey of the brain mechanisms underlying hemispheric asymmetries.5 Data in this table show that the model assuming a graded, right-hemispheric dominance in emotional face processing was supported by all experiments involving patients with unilateral brain damage and by most experiments involving healthy participants. In this last group of experiments, the valence hypothesis was supported by fewer than 20% of surveyed investigations, whereas all the other studies supported the right-hemisphere model more than the valence model. Consistent results with similar conclusions have been recently obtained by neurophysiological studies of the asymmetric processing of faces or pictures with different emotional valence.6–8 As with Lukito and colleagues’ findings,1 2 strongly negative (fearful and angry) expressions provoked extensive activation clusters in different cortical and subcortical structures, whereas no substantial activation was evoked by happy and sad faces, relative to neutral faces. A similar dissociation between negative and positive or neutral expressions was observed in studies investigating the capacity of the right and left hemisphere to identify positive and negative facial expressions.
Psychophysiological investigations of the lateralization of the autonomic components of emotions have also supported the right-hemisphere model of emotional laterality. This second line of research helps to clarify the specific negative emotions that have contributed to dissociation in the processing of emotional faces and the role of the right hemisphere in this dissociation. Studies in this second area of research have separately accounted for the psychophysiological correlates of emotional activation among patients with unilateral brain lesions, the hemispheric asymmetries in cardiac autonomic control, and the psychophysiological correlates of lateralized emotional stimulation among healthy participants. Most of these investigations documented a prevalent disruption of sympathetic functions in the right hemisphere and parasympathetic activities in the left hemisphere. Some studies, as reviewed by Hagemann and colleagues,9 documented an important reduction of galvanic skin responses and other indices of autonomic activation among patients with right brain lesions. Furthermore, they showed that large physiologic responses could be obtained after subliminal projection of emotional slides to the intact right hemisphere, whereas the vegetative response of patients with brain damage in the left hemisphere was much more variable. As for the hemispheric asymmetries in cardiac autonomic control, several investigations showed that unilateral brain lesions and unilateral pharmacological inactivations (achieved with the Wada test) shifted the cardiac autonomic balance toward a parasympathetic prevalence among patients with right-sided lesions and toward a sympathetic predominance among patients with lesions restricted to the left insula.10 Finally, Wittling and colleagues11,12 explored the psychophysiological correlates of selective emotional stimulation of the right and left hemisphere among healthy participants using different experimental procedures and obtained results indicative of differential specialization of the right hemisphere for sympathetic activities and of the left hemisphere for parasympathetic activities. Although not all studies have found a simple left–right dichotomy with respect to the laterality of the autonomic components of emotions, most have indicated a predominance of sympathetic functions in the right hemisphere and parasympathetic activities in the left hemisphere.
Given that the sympathetic system plays a critical role in the response to threatening events, the right hemisphere’s prevalence in sympathetic activities clearly suggests its prominent role in emotions related to threatening situations.
Reanalysis of meta-analysis data
With respect to the data reported by Lukito and colleagues, 1 activation clusters in the right and left cortical and subcortical structures, provoked by threatening faces, may not be necessarily be at odds with a model assuming a graded, right-hemisphere prevalence for threatening negative emotions. Relative to neutral faces, clusters provoked by fearful faces observed at the cortical level were mainly concerned with right-sided structures (i.e., the right inferior temporal gyrus, the right fusiform area, the right parahippocampal gyrus, and the right middle temporal gyrus), whereas left-sided structures (i.e., the left putamen and amygdala) were extensively activated by the same stimuli at the subcortical level only. Furthermore, within these subcortical structures, only the amygdala is well known for playing an important role in emotional processing (particularly of fear), whereas a similar function has rarely been attributed to the putamen. As the authors correctly acknowledge, the left-lateralized amygdala processing of emotional faces is in line with results from only 2 of 7 previous reviews on this subject; methodological reasons could explain the discrepancy between results. Indeed, Lukito and colleagues1 excluded studies involving the subliminal or subconscious perception of unattended emotional faces from their meta-analysis. However, Morris and colleagues13,14 showed that unconscious emotional learning can be mediated in the right hemisphere by a direct subcortical route connecting the thalamus with the amygdala, and, drawing on these pioneering studies, several others — excluded from the Lukito and colleagues’ 1 meta-analysis — have shown predominant activation of the right amygdala.15–17 For all these reasons, Lukito and colleagues’ findings do not necessarily disprove graded, right-hemisphere dominance for emotions.
A model of hemispheric asymmetries and threatening events
The third point that I would stress concerns the strong similarity that can be found between the claim that right-hemisphere dominance for emotions mainly concerns threatening events and the key points of an existing model of hemispheric asymmetries that assumes a graded, right-hemisphere dominance for emotions. This model distinguishes between an emotional system, mainly subsumed by the right hemisphere, and a cognitive system, mainly underpinned by the left hemisphere, as I have proposed elsewhere.10,18–20 Drawing on a model originally proposed by Oatley and Johnson-Laird,21 the hemispheric asymmetries model assumes that the emotional system may be considered an emergency system that is able to interrupt the ongoing action with an urgent procedure to rapidly select a new operative scheme. In contrast, the cognitive system is considered a more evolved and adaptive system, provided with a more propositional structure, that allows slower, controlled, and conscious processing of information and the development of appropriate operative strategies. The distinctive features of this emotional system could allow quick and automatic resolution of threatening events that can arise in a changing milieu; these features are, therefore, very appropriate to the model of emotions suggested by Lukito and colleagues. 1 On the other hand, some important features of the emotional system, such as its automatic and partly unconscious way of functioning, are shared with other aspects of the modus operandi typical of the right hemisphere. This can be exemplified by some well-known aspects of important right-hemispheric syndromes (e.g., unilateral spatial neglect, prosopagnosia, anosognosia) or by features of the normal functioning of the right side of the brain. In particular, a disruption of automatic components of behaviour is observed in unilateral spatial neglect and in prosopagnosia, whereas a partly unconscious way of functioning is observed in anosognosia and in the subconscious processing of emotions at the level of the right amygdala. It is well known that, in unilateral spatial neglect, the conscious and controlled (i.e., endogenous) components of attention are relatively spared, whereas the automatic (i.e., exogenous) components are much more impaired. 22–24 A similar disruption of the automatic components after damage to the right brain has been documented in prosopagnosia25,26 and in voice-recognition disorders.27,28
Discrepancies between results obtained by studies of emotional laterality based on functional neuroimaging and on neuropsychological investigations
Results of investigations on emotional laterality based on functional magnetic resonance imaging (fMRI) and meta-analyses are often at odds with those of neuropsychological and behavioural studies, as well as with results of other neuroimaging reviews. Three recent reviews of fMRI studies of emotional laterality had conflicting results.29–31 Morawetz and colleagues29 showed that 3 of 4 large-scale networks underlying perception, experience, and expression of emotion were left-lateralized. Stankovic30 suggested a right-hemispheric prevalence for emotional face perception and Palomero-Gallagher and Amunts31 proposed moving from models that supported an overall hemispheric asymmetry to models that incorporated multiple inter-related networks with different lateralization patterns for emotion. It is therefore possible that fMRI data may be better suited to evaluate the effect of a variable on the level or the intrahemispheric distribution of brain activation than to assess its possible hemispheric asymmetry. This could explain why Lukito and colleagues’ findings that brain activation was provoked more by threat-related emotional pictures than other facial expressions were more convincing than when they stated that their results failed to support the right-hemisphere and valence models of emotional laterality.
Shifting from these general thoughts to specific aspects of Lukito and colleagues’ meta-analysis,1 2 observations arise. For the sake of maximal homogeneity, the authors selected only empirical works that featured passive or active tasks, but did not specify what kinds of active tasks were involved in the studies included in the review. These studies may have entailed heterogeneous tasks, such as those concerning discrimination or classification of facial emotions, behavioural or psychophysiological reactions to threat-related stimuli, and so on. Furthermore, the authors did not account only for contrasts between emotional and neutral faces, but also contrasts between emotional face presentation and baseline activity, as this was the method used to measure activation contingent upon face-related tasks in a large number of the studies considered in their meta-analysis. Their analysis focused on the emotion versus neutral contrast, but data seem to show that emotion versus baseline comparisons were not completely identical to emotion versus neutral comparisons. A little extra caution in reconsidering these specific aspects of the meta-analysis by Lukito and colleagues1 could perhaps close the distance between their position and my position on hemispheric dominance for emotions.
On the basis of results obtained by the neuropsychological and psychophysiological research that have supported the right-hemisphere model of emotional laterality, reanalysis of meta-analysis data, the model of hemispheric asymmetries assuming that right-hemisphere dominance for emotions may mainly concern threatening events, and inconsistencies between results obtained by functional neuroimaging and neuropsychological investigations, I propose that emotions can be better classified on the basis of their relevance in threatening events than of the valence model or of the right-hemisphere dominance for emotions. I also propose that the new classification of emotions, based on the distinction between emotions related or unrelated to threatening events, can explain some relevant aspects of the model assuming a graded right-hemispheric prevalence of emotions.
Footnotes
Competing interests: None declared.
- Received January 5, 2024.
- Revision received April 2, 2023.
- Accepted April 29, 2024.
This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) licence, which permits use, distribution and reproduction in any medium, provided that the original publication is properly cited, the use is noncommercial (i.e., research or educational use), and no modifications or adaptations are made. See: https://creativecommons.org/licenses/by-nc-nd/4.0/