Affective neuroscience


Affective neuroscience is the study of the neural mechanisms of emotion. This interdisciplinary field combines neuroscience with the psychological study of personality, emotion, and mood. The putative existence of 'basic emotions' and their defining attributes represents a long lasting and yet unsettled issue in psychology.
Affective neuroscience developed at a time when cognitive neuroscience focused on non-emotional cognition, such as attention or memory. It might be called the cognitive neuroscience of emotions.
However, when reading this article, please note that there are few references to newer studies occurring after 2012 that have changed many of the perspectives in this article. For example, the section relating to emotions and the right hemisphere refer mainly to studies prior to 2006, and these models have been abandoned in favor of the new evidence showing interhemispheric connectivity and the role that the salience network plays in interpreting instinctual emotional responses to the world.

Brain areas related to emotion

Emotions are thought to be related to activity in brain areas that direct our attention, motivate our behavior, and determine the significance of what is going on around us. Pioneering work by Paul Broca, James Papez, and Paul D. MacLean suggested that emotion is related to a group of structures in the center of the brain called the limbic system, which includes the hypothalamus, cingulate cortex, hippocampi, and other structures. Research has shown that limbic structures are directly related to emotion, but non-limbic structures have been found to be of greater emotional relevance. The following brain structures are currently thought to be involved in emotion:

Main structures of the limbic system

The right hemisphere has been proposed over time as being directly involved in the processing of emotion. Scientific theory regarding the role of the right hemisphere has developed over time and resulted in several models of emotional functioning. C.K. Mills was one of the first researchers to propose a direct link between the right hemisphere and emotional processing, having observed decreased emotional processing in patients with lesions to the right hemisphere. Emotion was originally thought to be processed in the limbic system structures such as the hypothalamus and amygdala. As of the late 1980s to early 1990s however, neocortical structures were shown to have an involvement in emotion. These findings led to the development of the right hemisphere hypothesis and the valence hypothesis.

The right hemisphere hypothesis

The right hemisphere hypothesis asserts that the right hemisphere of the neocortical structures is specialized for the expression and perception of emotion. The Right hemisphere has been linked with mental strategies that are nonverbal, synthetic, integrative, holistic, and Gestalt which makes it ideal for processing emotion. The right hemisphere is more in touch with subcortical systems of autonomic arousal and attention as demonstrated in patients that have increased spatial neglect when damage is associated to the right brain as opposed to the left brain. Right hemisphere pathologies have also been linked with abnormal patterns of autonomic nervous system responses. These findings would help signify the relationship of the subcortical brain regions to the right hemisphere as having a strong connection.

The valence hypothesis

The valence hypothesis acknowledges the right hemisphere's role in emotion, but asserts that it is mainly focused on the processing of negative emotions whereas the left hemisphere processes positive emotions. The mode of processing of the two hemispheres has been the discussion of much debate. One version suggests the lack of a specific mode of processes, stating that the right hemisphere is solely negative emotion and the left brain is solely positive emotion. A second version suggests that there is a complex mode of processing that occurs, specifically that there is a hemispheric specialization for the expressing and experiencing of emotion, with the right hemisphere predominating in the experiencing of both positive and negative emotion. More recently, the frontal lobe has been the focus of a large amount of research, stating that the frontal lobes of both hemispheres are involved in the emotional state, while the right posterior hemisphere, the parietal and temporal lobes, is involved in the processing of emotion. Decreased right parietal lobe activity has been associated with depression and increased right parietal lobe activity with anxiety arousal. The increasing understanding of the role the different hemispheres play has led to increasingly complicated models, all based some way on the original valence model.

Relationship to cognitive neuroscience

Despite their interactions, the study of cognition has, until the late 1990s, excluded emotion and focused on non-emotional processes. As a result, the study of the neural basis of non-emotional and emotional processes emerged as two separate fields: cognitive neuroscience and affective neuroscience. The distinction between non-emotional and emotional processes is now thought to be largely artificial, as the two types of processes often involve overlapping neural and mental mechanisms. Thus, when cognition is taken at its broadest definition, affective neuroscience could also be called the cognitive neuroscience of emotion.

Cognitive neuroscience tasks in affective neuroscience research

Emotion Go/No-Go

The emotion go/no-go task has been frequently used to study behavioral inhibition, particularly emotional modulation of this inhibition. A derivation of the original go/no-go paradigm, this task involves a combination of affective “go cues”, where the participant must make a motor response as quickly as possible, and affective “no-go cues,” where a response must be withheld. Because “go cues” are more common, the task is able to measure one's ability to inhibit a response under different emotional conditions.
The task is common in tests of emotion regulation, and is often paired with neuroimaging measures to localize relevant brain function in both healthy individuals and those with affective disorders. For example, go/no-go studies converge with other methodology to implicate areas of the prefrontal cortex during inhibition of emotionally valenced stimuli.

Emotional Stroop

The emotional Stroop task, an adaptation to the original Stroop, measures attentional bias to emotional stimuli. Participants must name the ink color of presented words while ignoring the words themselves. In general, participants have more difficulty detaching attention from affectively valenced words, than neutral words. This interference from valenced words is measured by the response latency in naming the color of neutral words as compared with emotional words.
This task has been often used to test selective attention to threatening and other negatively valenced stimuli, most often in relation to psychopathology. Disorder specific attentional biases have been found for a variety of mental disorders. For example, participants with spider phobia show a bias to spider-related words but not other negatively valenced words. Similar findings have been attributed to threat words related to other anxiety disorders. However, other studies have questioned these findings. In fact, anxious participants in some studies show the Stroop interference effect for both negative and positive words, when the words are matched for emotionality. This means that the specificity effects for various disorders may be largely attributable to the semantic relation of the words to the concerns of the disorder, rather than simply the emotionality of the words.

Ekman 60 faces task

The Ekman faces task is used to measure emotion recognition of six basic emotions. Black and white photographs of 10 actors are presented, with each actor displaying each basic emotion. Participants are usually asked to respond quickly with the name of the displayed emotion. The task is a common tool to study deficits in emotion regulation in patients with dementia, Parkinson's, and other cognitively degenerative disorders. However, the task has also been used to analyze recognition errors in disorders such as borderline personality disorder, schizophrenia, and bipolar disorder.

Dot probe (emotion)

The emotional dot-probe paradigm is a task used to assess selective visual attention to and failure to detach attention from affective stimuli. The paradigm begins with a fixation cross at the center of a screen. An emotional stimulus and a neutral stimulus appear side by side, after which a dot appears behind either the neutral stimulus or the affective stimulus. Participants are asked to indicate when they see this dot, and response latency is measured. Dots that appear on the same side of the screen as the image the participant was looking at will be identified more quickly. Thus, it is possible to discern which object the participant was attending to by subtracting the reaction time to respond to congruent versus incongruent trials.
The best documented research with the dot probe paradigm involves attention to threat related stimuli, such as fearful faces, in individuals with anxiety disorders. Anxious individuals tend to respond more quickly to congruent trials, which may indicate vigilance to threat and/or failure to detach attention from threatening stimuli. A specificity effect of attention has also been noted, with individuals attending selectively to threats related to their particular disorder. For example, those with social phobia selectively attend to social threats but not physical threats. However, this specificity may be even more nuanced. Participants with obsessive-compulsive disorder symptoms initially show attentional bias to compulsive threat, but this bias is attenuated in later trials due to habituation to the threat stimuli.

Fear potentiated startle

has been utilized as a psychophysiological index of fear reaction in both animals and humans. FPS is most often assessed through the magnitude of the eyeblink startle reflex, which can be measured by electromyography. This eyeblink reflex is an automatic defensive reaction to an abrupt elicitor, making it an objective indicator of fear. Typical FPS paradigms involve bursts of noise or abrupt flashes of light transmitted while an individual attends to a set of stimuli. Startle reflexes have been shown to be modulated by emotion. For example, healthy participants tend to show enhanced startle responses while viewing negatively valenced images and attenuated startle while viewing positively valenced images, as compared with neutral images.
The startle response to a particular stimulus is greater under conditions of threat. A common example given to indicate this phenomenon is that one's startle response to a flash of light will be greater when walking in a dangerous neighborhood at night than it would under safer conditions. In laboratory studies, the threat of receiving shock is enough to potentiate startle, even without any actual shock.
Fear potentiated startle paradigms are often used to study fear learning and extinction in individuals with posttraumatic stress disorder and other anxiety disorders. In fear conditioning studies, an initially neutral stimulus is repeatedly paired with an aversive one, borrowing from classical conditioning. FPS studies have demonstrated that PTSD patients have enhanced startle responses during both danger cues and neutral/safety cues as compared with healthy participants.

Learning

There are many ways affect plays a role during learning. Recently, affective neuroscience has done much to discover this role. Deep, emotional attachment to a subject area allows a deeper understanding of the material and therefore, learning occurs and lasts. When reading, the emotions one is feeling in comparison to the emotions being portrayed in the content affects ones comprehension. Someone who is feeling sad will understand a sad passage better than someone feeling happy. Therefore, a student's emotion plays a big role during the learning process.
Emotion can also be embodied or perceived from words read on a page or a person's facial expression. Neuroimaging studies using fMRI have demonstrated that the same area of the brain being activated when one is feeling disgust is also activated when one observes another person feeling disgust. In a traditional learning environment, the teacher's facial expression can play a critical role in students' language acquisition. Showing a fearful facial expression when reading passages that contain fearful tones facilitates students learning of the meaning of certain vocabulary words and comprehension of the passage.

Meta-analyses

A meta-analysis is a statistical approach to synthesizing results across multiple studies. Several meta-analyses examining the brain basis of emotion have been conducted. In each meta-analysis, studies were included that investigate healthy, unmedicated adults and that used subtraction analysis to examine the areas of the brain that were more active during emotional processing than during a neutral control condition. The meta-analyses to date predominantly focus on two theoretical approaches, locationist approaches and psychological construction approaches.
It is being debated regarding the existence of the neurobiological basis of emotion. The existence of so-called 'basic emotions' and their defining attributes represents a long lasting and yet unsettled issue in psychology. The available research suggests that the neurobiological existence of basic emotions is still tenable and heuristically seminal, pending some reformulation.

Locationist approaches

These approaches to emotion hypothesize that several emotion categories are biologically basic. In this view, emotions are inherited biologically based modules that cannot be broken down into more basic psychological components. Models following a locationist approach to emotion hypothesize that all mental states belonging to a single emotional category can be consistently and specifically localized to either a distinct brain region or a defined networks of brain regions. Each basic emotion category also shares other universal characteristics: distinct facial behavior, physiology, subjective experience and accompanying thoughts and memories.

Psychological constructionist approaches

This approach to emotion hypothesizes that emotions like happiness, sadness, fear, anger and disgust are constructed mental states that occur when many different systems in the brain work together. In this view, networks of brain regions underlie psychological operations that interact to produce many different kinds of emotion, perception, and cognition. One psychological operation critical for emotion is the network of brain regions that underlie valence and arousal. Emotions emerge when neural systems underlying different psychological operations interact, producing distributed patterns of activation across the brain. Because emotions emerge from more basic components, there is heterogeneity within each emotion category; for example, a person can experience many different kinds of fear, which feel differently, and which correspond to different neural patterns in the brain. Thus, this view presents a different approach to understanding the neural bases of emotion than locationist approaches.

Phan et al. 2002

In the first neuroimaging meta-analysis of emotion, Phan et al. analyzed the results of 55 studies published in peer reviewed journal articles between January 1990 and December 2000 to determine if the emotions of fear, sadness, disgust, anger, and happiness were consistently associated with activity in specific brain regions. All studies used fMRI or PET techniques to investigate higher-order mental processing of emotion. The authors’ analysis approach was to tabulate the number of studies that reported activation in specific brain regions during tasks inducing fear, sadness, disgust, anger, and happiness. For each brain region, statistical chi-squared analysis was conducted to determine if the proportion of studies reporting activation during one emotion was significantly higher than the proportion of studies reporting activation during the other emotions. Two regions showed this statistically significant pattern across studies. In the amygdala, 66% of studies inducing fear reported activity in this region, as compared to ~20% of studies inducing happiness, ~15% of studies inducing sadness. In the subcallosal cingulate, 46% of studies inducing sadness reported activity in this region, as compared to ~20% inducing happiness and ~20% inducing anger. This pattern of clear discriminability between emotion categories was in fact rare, with a number of other patterns occurring in limbic regions, paralimbic regions, and uni/heteromodal regions. Brain regions implicated across discrete emotion included the basal ganglia and medial prefrontal cortex.

Murphy et al. 2003

Murphy, et al. 2003 analyzed 106 peer reviewed journals published between January 1994 and December 2001 to examine the evidence for regional specialization of discrete emotions across a larger set of studies that Phan et al. Studies included in the meta-analysis measured activity in the whole brain and regions of interest. 3-D Kolmogorov-Smirnov statistics were used to compare rough spatial distributions of 3-D activation patterns to determine if statistically significant activations were specific to particular brain regions for all emotional categories. This pattern of consistently activated, regionally specific activations was identified in four brain regions: amygdala with fear, insula with disgust, globus pallidus with disgust, and lateral orbitofrontal cortex with anger. The amygdala was consistently activated in ~40% of studies inducing fear, as compared to less than 20% studies inducing happiness, sadness, or anger. The insula was consistently activated in ~ 70% of studies inducing disgust, as compared to sadness, anger, fear, and happiness. Similar to the insula, the globus pallidus was consistently activated in ~70% of studies inducing disgust, as compared to less than 25% of studies inducing sadness, fear, anger or happiness. The lateral orbitofrontal cortex was consistently activated in over 80% of studies inducing anger, as compared to fear, sadness, happiness and disgust. Other regions showed different patterns of activation across categories. For example, both the dorsal medial prefrontal cortex and the rostral anterior cingulate cortex showed consistent activity across emotions.

Barrett et al. 2006

Barrett, et al. 2006 examined 161 studies published between 1990-2001, subsets of which were analyzed in previous meta-analyses. In this review, the authors examined the locationist hypothesis by comparing the consistency and specificity of prior meta-analytic findings specific to each hypothesized basic emotion. Consistent neural patterns were defined by brain regions showing increased activity for a specific emotion, regardless of the method of induction used. Specific neural patterns were defined as architecturally separate circuits for one emotion vs. the other emotions. In general, the results supported consistency among the findings of Phan et al. and Murphy et al., but not specificity. Consistency was determined through the comparison of chi-squared analyses that revealed whether the proportion of studies reporting activation during one emotion was significantly higher than the proportion of studies reporting activation during the other emotions. Specificity was determined through the comparison of emotion-category brain-localizations by contrasting activations in key regions that were specific to particular emotions. Increased amygdala activation during fear was the most consistently reported across induction methods. Both meta-analyses also reported increased activations in regions of the anterior cingulate cortex during sadness, although this finding was less consistent and was not specific to sadness. Both meta-analyses also found that disgust was associated with increased activity in the basal ganglia, but these findings were neither consistent nor specific. Neither consistent nor specific activity was observed across the meta-analyses for anger or for happiness. This meta-analysis additionally introduced the concept of the basic, irreducible elements of emotional life as dimensions such as approach and avoidance. This dimensional approach involved in psychological constructionist approaches is further examined in later meta-analyses of Kober et al. 2008 and Lindquist et al. 2012.

Kober et al. 2008

Instead of investigating specific emotions, Kober, et al. 2008 reviewed 162 neuroimaging studies published between 1990-2005 to determine if groups of brain regions show consistent patterns of activation during emotional experience and during emotion perception. This meta-analysis used multilevel kernal density analysis to examine fMRI and PET studies, a technique that prevents single studies from dominating the results and that enables studies with large sample sizes to exert more influence upon the results. MKDA was used to establish a neural reference space that includes the set of regions showing consistent increases across all studies. Next, this neural reference space was partitioned into functional groups of brain regions showing similar activation patterns across studies by first using multivariate techniques to determine co-activation patterns and then using data-reduction techniques to define the functional groupings. Consistent with a psychological construction approach to emotion, the authors discuss each functional group in terms more basic psychological operations. The first “Core Limbic” group included the left amygdala, hypothalamus, periaqueductal gray/thalamus regions, and amygdala/ventral striatum/ventral globus pallidus/thalamus regions, which the authors discuss as an integrative emotional center that plays a general role in evaluating affective significance. The second “Lateral Paralimbic” group included the ventral anterior insula/frontal operculum/right temporal pole/ posterior orbitofrontal cortex, the anterior insula/ posterior orbitofrontal cortex, the ventral anterior insula/ temporal cortex/ orbitofrontal cortex junction, the midinsula/ dorsal putamen, and the ventral striatum /mid insula/ left hippocampus, which the authors suggest plays a role in motivation, contributing to the general valuation of stimuli and particularly in reward. The third “Medial Prefrontal Cortex” group included the dorsal medial prefrontal cortex, pregenual anterior cingulate cortex, and rostral dorsal anterior cingulate cortex, which the authors discuss as playing a role in both the generation and regulation of emotion. The fourth “Cognitive/ Motor Network” group included right frontal operculum, the right interior frontal gyrus, and the pre-supplementray motor area/ left interior frontal gyrus, regions that are not specific to emotion, but instead appear to play a more general role in information processing and cognitive control. The fifth “Occipital/ Visual Association” group included areas V8 and V4 of the primary visual cortex, the medial temporal lobe, and the lateral occipital cortex, and the sixth “Medial Posterior” group included posterior cingulate cortex and area V1 of the primary visual cortex. The authors suggest that these regions play a joint role in visual processing and attention to emotional stimuli.

Vytal et al. 2010

Vytal, et al. 2010 examined 83 neuroimaging studies published between 1993-2008 to examine whether neuroimaging evidence supports the idea of biologically discrete, basic emotions. Consistency analyses identified brain regions that were associated with a given emotion. Discriminability analyses identified brain regions that were significantly, differentially active when contrasting pairs of discrete emotions. This meta-analysis examined PET or fMRI studies that reported whole brain analyses identifying significant activations for at least one of the five emotions relative to a neutral or control condition. The authors used activation likelihood estimation to perform spatially sensitive, voxel-wise statistical comparisons across studies. This technique allows for direct statistical comparison between activation maps associated with each discrete emotion. Thus, discriminability between the five discrete emotion categories was assessed on a more precise spatial scale than what had been accomplished in prior meta-analyses. Consistency was first assessed by comparing the ALE map generated across studies for each emotion to ALE map generated by random permutations. Discriminability was then assessed by pair-wise contrasts of individual emotion ALE maps across all basic emotions pairings. Consistent and discriminable patterns of neural activation were observed for the five emotional categories. Happiness was consistently associated with activity in 9 regional brain clusters, the largest located in the right superior temporal gyrus. For the first time, happiness was discriminated from the other emotional categories, with the largest clusters of activity specific to happiness located in right superior temporal gyrus and left rostral anterior cingulate cortex. Sadness was consistently associated with 35 clusters and was discriminated from the other emotion categories by significantly greater activity in left medial frontal gyrus, right middle temporal gyrus, and right inferior frontal gyrus. Anger was consistently associated with activity in 13 clusters, and was discriminated from the other emotion categories by significantly greater activity in bilateral inferior frontal gyrus, and in right parahippocampal gyrus. Fear was consistently associated with 11 clusters and was discriminated from the other emotion categories by significantly greater activity in the left amygdala and left putamen. Disgust was consistently activated with 16 clusters and was discriminated from the other emotion categories by significantly greater activity in the right putamen and the left insula.

Lindquist et al. 2012

Lindquist, et al. 2012 reviewed 91 PET and fMRI studies published between January 1990 and December 2007. The studies included in this meta-analysis used induction methods that elicit emotion experience or emotion perception of fear, sadness, disgust, anger, and happiness. The goal was to compare locationist approaches with psychological constructionist approaches to emotion. Similar to Kober et al. described above, a Multilevel Peak Kernel Density Analysis transformed the individual peak activations reported across study contrasts into a neural reference space. The density analysis was then used to identify voxels within the neural reference space with more consistent activations for a specific emotion category than all other emotions. Chi-squared analysis was used to create statistical maps that indicated if each previously identified and consistently active regions were more frequently activated in studies of each emotion category versus the average of all other emotions, regardless of activations elsewhere in the brain. Chi-squared analysis and density analysis both defined functionally consistent and selective regions, or regions which showed a relatively more consistent increase in activity for the experience or perception of one emotion category across studies in the literature. Thus, a selective region could present increased activations relatively more so to one emotion category while also having a response to multiple other emotional categories. A series of logistic regressions were then performed to identify if any of the regions that were identified as consistent and selective to an emotion category were additionally specific to a given category. Regions were defined as specific to a given emotion if they showed increased activations for only one emotional category, and never showed increased activity during instances of the other emotional categories. In other words, a region could be defined as consistent, selective and specific for e.g. fear perception if it only showed significantly greater increases in activation during the perception of fear and did not show increased activity during any other emotion categories. However, the same region would be defined as only consistent and selective to fear perception if it additionally displayed increased activations during anger perception. Strong support for the locationist approach was defined as evidence that basic emotion categories consistently map onto areas of the brain that specifically activate in response to instances of only one emotional category. Strong support for the constructionist approach was defined as evidence that multiple psychological operations consistently occur across many brain regions and multiple emotional categories.
The results indicated that many brain regions demonstrated consistent and selective activations in the experience or perception of an emotion category. Consistent with constructionist models, however, no region demonstrated functional specificity for the emotions of fear, disgust, happiness, sadness or anger. Based on the existing scientific literature, the authors proposed different roles for the brain regions that have traditionally been associated with only one emotion category. The authors propose that the amygdala, anterior insula, orbitofrontal cortex each contribute to “core affect,” which are basic feelings that are pleasant or unpleasant with some level of arousal. The amygdala, for example, appears to play a more general role in indicating if external sensory information is motivationally salient, and is particularly active when a stimulus is novel or evokes uncertainty. The anterior insula may represent core affective feelings in awareness across a number of emotion categories, driven largely by sensations originating in the body. The orbitofrontal cortex appears to function as a site for integrating sensory information from the body and sensory information from the world to guide behavior. Closely related to core affect, the authors propose that anterior cingulate and dorsolateral prefrontal cortex play vital roles in attention, with anterior cingulate supporting the use of sensory information for directing attention and motor responses during response selection and with dorsolateral prefrontal cortex supporting executive attention. In many psychological construction approaches, emotions also involve the act of interpreting one's situation in the world relative to the internal state of the body, or what is referred to as “conceptualization.” In support of this idea, the dorsomedial prefrontal cortex and hippocampus were consistently active in this meta-analysis, regions that appear to play an important role conceptualizing during emotion, which are also involved in simulating previous experience. Language is also central to conceptualizing, and regions that support language, including ventrolateral prefrontal cortex, were also consistently active across studies of emotion experience and perception.